View Full Version : Better iPod 5.5 profile for TV viewing?
simonhowson
15th April 2008, 14:59
I want to know how to improve the quality encoding when viewing videos from an iPod Classic on a 76 cm widescreen TV? I want settings that will be better than the iPod 5.5 profile in MeGUI.
I am using 640 wide, with the vertical resolution dependent on the aspect ratio.
The MeGUI 5.5 profile seems to restrict the maximum bitrate to 1500, even though Apple's specs for the iPod Classic (http://www.apple.com/ipodclassic/specs.html) say it can playback
H.264 video, up to 2.5 Mbps, 640 by 480 pixels, 30 frames per second, Baseline Profile up to Level 3.0
So I've increased the "VBV Maximum Bitrate" to 2500.
Next I thought that perhaps CONSTANT QUALITY may help improve quality, compared to the iPod 5.5 profile's setting of Automated 2pass, targeting 1000 bps. I started at 15, which produced 1.5 GB files from a 110 minute, 4:3 film. The quality was great, but this is too big. It seems that 20 is a reasonable setting, but this seems to result in 700 - 800 MB files for 2.35:1 films, and 1 - 1.1 GB files for 4:3 films.
Does it make sense using Constant Quality mode, if the bitrate is constrained to 2500 max? Would Automated 2 pass produce better quality than such a constrained Constant Quality mode?
Should I consider using anamorphic encoding, considering that the image has to be zoomed to fill the TV screen? (for widescreen films I mean)
I have turned the deblocking filter on, because it seems to improve quality quite a lot. Does it make sense using a setting of 1:1 when the file has to be zoomed to fill the widescreen TV, or should I just leave it at 0:0?
Is there a way to make files that are both iPod Classic compatible, and DxVA compatible? Or is it unnecessary because iPod compatible files don't require that much processing power anyway?
Lastly! What's the easiest way to learn about filtering? I basically just use the default noise filter in MeGUI (undot). But is there a way that the clip can be analysed, and have the filtering applied automatically?
informationspy
15th April 2008, 16:10
Get A Very Good Encoding Software And Use Advace Function. And Set The Pixes By Pixes You Wanted And You Still Can Set The Quality Of Pictures .
That Will Ensure You Can Get What You Wanted !
Ranguvar
15th April 2008, 17:09
No idea what informationspy just said (and for the record, it's gonna be hard to find something that can outdo x264).
Anyways, CQ usually doesn't do as well as CRF for an actual encode. And in that situation, I'd say CRF > 2-pass. But you'd constrain 2-pass too... 2-pass is VBR, and you don't want the bitrate to peak over 2.5Mb/s.
For filtering, check out www.avisynth.org and the AviSynth forum here. It's not that hard, and you'll learn new tricks to optimize your video. UnDot is an extremely light denoiser that's good at slight mosquito noise and invisible noise.
CQ or CRF at anything less than 18 is rather ridiculous, IMO. 18 should be completely transparent usually. 20-24 are what you'll probably be looking at.
If you increase deblocking, you'll get less blocks, but less details. It will NOT get rid of blocks that will appear if the image is being stretched.
Your statement on anamorphic doesn't make sense to me...
And iPod videos shouldn't be that hard to decode on most machines, no.
simonhowson
15th April 2008, 17:52
For filtering, check out www.avisynth.org and the AviSynth forum here. It's not that hard, and you'll learn new tricks to optimize your video. UnDot is an extremely light denoiser that's good at slight mosquito noise and invisible noise.
I tend to encode a lot of films from the 1940s and 1950s. They tend to be quite grainy because of the old slow film stocks. Can you recommend a filter for quite grainy footage? Having said that, I don't want to remove all the grain, because then it would look like video.
CQ or CRF at anything less than 18 is rather ridiculous, IMO. 18 should be completely transparent usually. 20-24 are what you'll probably be looking at.
20 looked good. But, as you suggested, it seems to me that using Constant Quality is pointless when you have to enforce a 2500 limit on the bitrate.
Does that limit mean that it would just be impossible to make all videos look good, because 640 X 480 resolution occasionally will need more than 2500, unless the filtering is done perfect?
If you increase deblocking, you'll get less blocks, but less details. It will NOT get rid of blocks that will appear if the image is being stretched.
Yeah, I think it is worth leaving on (the iPod Classic has a pretty high capacity battery), but I think I'll leave it at 0:0. My motto is when in doubt, leave it at the defaults
Your statement on anamorphic doesn't make sense to me...
I plug the iPod Classic into my TV using an S-Video cable. It is a widescreen TV, but once I start playing the video it appears in the middle of the screen. For a 4:3 film this is fine. But for widescreen films, I press the ZOOM button on my TV remote so the widescreen image touches the sides of the screen.
You know it is like if you have a non-anamorphic encoded DVD. On a widescreen TV you will have to zoom in so it fits the screen. This process of zooming degrades the image somewhat, because it is spreading the lower resolution image over a larger screen area.
Of course I could do something tricky, and figure out how to encode a distorted iPod video. But then I wouldn't be able to watch it properly on the iPod itself!
Thanks for your reply!
Ranguvar
15th April 2008, 21:27
I tend to encode a lot of films from the 1940s and 1950s. They tend to be quite grainy because of the old slow film stocks. Can you recommend a filter for quite grainy footage? Having said that, I don't want to remove all the grain, because then it would look like video.There are many degrainers out there, you just need to search. MVDegrain and TemporalDegrain are excellent filters, but you need to learn a bit about AviSynth first.
20 looked good. But, as you suggested, it seems to me that using Constant Quality is pointless when you have to enforce a 2500 limit on the bitrate.I didn't say that. 2500 is a fine max, increasing that max won't really help. What I said was that 2-pass, CQ, and CRF would all be affected around the same by that max. I recommend CRF over CQ because it is more efficient for a real encode, IMO.
IDoes that limit mean that it would just be impossible to make all videos look good, because 640 X 480 resolution occasionally will need more than 2500, unless the filtering is done perfect?Nope. A CBR 2500kb/s encode would look transparent (same as source) in most cases, with good encode settings.
IYeah, I think it is worth leaving on (the iPod Classic has a pretty high capacity battery), but I think I'll leave it at 0:0. My motto is when in doubt, leave it at the defaultsDeblocking really helps in most cases. Decrease the values for more detail if you have a high enough bitrate to avoid blocking, increase them to deal with excessive blocking. When in doubt, try different values and compare ;)
I plug the iPod Classic into my TV using an S-Video cable. It is a widescreen TV, but once I start playing the video it appears in the middle of the screen. For a 4:3 film this is fine. But for widescreen films, I press the ZOOM button on my TV remote so the widescreen image touches the sides of the screen.
You know it is like if you have a non-anamorphic encoded DVD. On a widescreen TV you will have to zoom in so it fits the screen. This process of zooming degrades the image somewhat, because it is spreading the lower resolution image over a larger screen area.!What resolution are you encoding to? And what are you coming from?
simonhowson
16th April 2008, 02:26
What resolution are you encoding to? And what are you coming from?
My source are DVDs, I'm resizing down to a width of 640, and a vertical resolution dependent on the aspect ratio.
I have tried encoding anamorphic 640 x X, but I think it degrades the quality.
At the moment I am trying for a target bitrate of 1100, rather than 1000, deblocking 1:1, and the max at 2500, rather than 1500.
Over all I prefer a smoother image. When sitting up close to the TV I can see the difference, but when I sit in my usual position it looks good.
Ranguvar
16th April 2008, 04:58
Anamorphic should not degrade the image quality; if anything it should be higher than with fixed resize. For example, a fixed 640x360 is 16:9. However, if you did an anamorphic 640x480 encode, you would have a squeezed fullscreen image which would be stretched to 853x480, which is a higher resolution. Now, anamorphic 640x480 to 853x480 is not as high quality as pure 853x480 (of course you couldn't use an odd resolution for a normal resize anyways), but it's better than 640x360. And 720x480 anamorphic to 853x480 would be better than 640x480 anamorphic, because it has to stretch less.
simonhowson
16th April 2008, 07:00
Anamorphic should not degrade the image quality; if anything it should be higher than with fixed resize. For example, a fixed 640x360 is 16:9. However, if you did an anamorphic 640x480 encode, you would have a squeezed fullscreen image which would be stretched to 853x480, which is a higher resolution. Now, anamorphic 640x480 to 853x480 is not as high quality as pure 853x480 (of course you couldn't use an odd resolution for a normal resize anyways), but it's better than 640x360. And 720x480 anamorphic to 853x480 would be better than 640x480 anamorphic, because it has to stretch less.
OK I will experiment with anamorphic. But it seemed to me that the quality was going down, because there are more bits to encode? Using Constant Quality 20 the anamorphic files were 20 - 30% bigger than the Quality 20 fixed resize files.
Now that I have increased the maximum bitrate from 1500 to 2500 I am getting a strange "VBV underflow" error during my encodes. Is this because I need to increase the size of the VBV buffer? Here is an example of the error:
x264 [warning]: VBV underflow (-10488 bits)
x264 [warning]: VBV underflow (-4400 bits)
x264 [warning]: VBV underflow (-13496 bits)
x264 [warning]: VBV underflow (-15680 bits)
x264 [warning]: VBV underflow (-6464 bits)
x264 [warning]: VBV underflow (-10584 bits)
x264 [warning]: VBV underflow (-10328 bits)
x264 [warning]: VBV underflow (-3032 bits)
x264 [warning]: VBV underflow (-25664 bits)
x264 [warning]: VBV underflow (-10248 bits)
x264 [warning]: VBV underflow (-2240 bits)
x264 [warning]: VBV underflow (-11616 bits)
x264 [warning]: VBV underflow (-14776 bits)
x264 [warning]: VBV underflow (-680 bits)
x264 [warning]: VBV underflow (-2184 bits)
x264 [warning]: VBV underflow (-5128 bits)
x264 [warning]: VBV underflow (-17040 bits)
x264 [warning]: VBV underflow (-3152 bits)
x264 [warning]: VBV underflow (-25152 bits)
x264 [warning]: VBV underflow (-5440 bits)
x264 [warning]: VBV underflow (-2288 bits)
x264 [warning]: VBV underflow (-4208 bits)
x264 [warning]: VBV underflow (-1512 bits)
x264 [warning]: VBV underflow (-1928 bits)
x264 [warning]: VBV underflow (-14736 bits)
x264 [warning]: VBV underflow (-11296 bits)
x264 [warning]: VBV underflow (-4688 bits)
x264 [warning]: VBV underflow (-20520 bits)
x264 [warning]: VBV underflow (-36160 bits)
x264 [warning]: VBV underflow (-22656 bits)
x264 [warning]: VBV underflow (-8272 bits)
x264 [warning]: VBV underflow (-1600 bits)
x264 [warning]: VBV underflow (-26896 bits)
x264 [warning]: VBV underflow (-28704 bits)
x264 [warning]: VBV underflow (-22272 bits)
x264 [warning]: VBV underflow (-35032 bits)
x264 [warning]: VBV underflow (-2848 bits)
x264 [warning]: VBV underflow (-7304 bits)
x264 [warning]: VBV underflow (-5512 bits)
x264 [warning]: VBV underflow (-4744 bits)
x264 [warning]: VBV underflow (-6304 bits)
x264 [warning]: VBV underflow (-15968 bits)
x264 [warning]: VBV underflow (-11992 bits)
x264 [warning]: VBV underflow (-71104 bits)
x264 [warning]: VBV underflow (-7280 bits)
x264 [warning]: VBV underflow (-7264 bits)
Now that VBV max is 2500, should VBV buffer be higher?
I am encoding using Automatted 2pass, with the bitrate set to 1000. Should the buffer be bigger than the target bitrate?
Southstorm
16th April 2008, 13:36
iPod Classic Specs: H.264 video, up to 2.5 Mbps, 640 by 480 pixels, 30 frames per second, Baseline Profile up to Level 3.0 with AAC-LC audio up to 160 Kbps, 48kHz, stereo audio in .m4v, .mp4, and .mov file formats;
Your iPod Classic supports a Level 3.0 profile. The VBV buffer, and VBV maxrate can be as high as 10000 KB/s even though your average bitrate is only 2500 KB/s.
Your errors are coming from the low VBV maxrate settings you're using at 2500 KB/s.
simonhowson
17th April 2008, 17:25
iPod Classic Specs:
Your iPod Classic supports a Level 3.0 profile. The VBV buffer, and VBV maxrate can be as high as 10000 KB/s even though your average bitrate is only 2500 KB/s.
Your errors are coming from the low VBV maxrate settings you're using at 2500 KB/s.
I'm using 1000 as the average bitrate, 2500 is the maximum bitrate, because according to Apple's specs that is the most it will reliably run on.
It will accept files with bitrates that exceed 2500, but if the file is encoded with the deblocking filter, the files stutter during playback. It only happens briefly during very complicated scenes, but it is annoying, so I want to avoid it.
What should I set the VBV buffer to? Somewhere between 1000 and 2500? What about 1750?
Southstorm
17th April 2008, 17:40
Try a few settings and experiment...The point is that a Level 3.0 profile requires the device to handle a bitrate "spike" every now and then up to 10000 KB/s.
I have a 5G iPod that only handles Level 1.3 profile, which means 750 Kb/s max bitrate, but a VBV-buffer up to 2000 KB/s. I use these settings all the time succesfully.
Ranguvar
17th April 2008, 20:00
Southstorm, the 5G and 5.5G both have the same video support. I believe there is something different in the way they do deblocking (I think the 5G doesn't do it to save cycles). The max settings you described apply only to QVGA video, and for both of them.
Southstorm
17th April 2008, 20:06
@Ranguvar
You are correct in part, however, there were two 5G iPod releases. A late 2005 release, and a mid 2006 release. The earlier of the two, which is what i have, has these specs: From the Apple Site: H.264 video: up to 768 Kbps, 320 x 240, 30 frames per sec., Baseline Profile up to Level 1.3 with AAC-LC up to 160 Kbps, 48 Khz, stereo audio in .m4v, .mp4 and .mov file formats
MPEG-4 video: up to 2.5 mbps, 480 x 480, 30 frames per sec., Simple Profile with AAC-LC up to 160 Kbps, 48 Khz, stereo audio in .m4v, .mp4 and .mov file formats
Southstorm
17th April 2008, 23:06
The options in this x264 commandline should work for the 5.5G iPod and look decent on TV as well. --keyint 250 --min-keyint 25 --scenecut 40 --bframes 0 --no-b-adapt --no-cabac --ref 1
--bitrate 1500 --vbv-maxrate 10000 --vbv-bufsize 10000 --partitions p8x8,p4x4,b8x8,i4x4
--direct-8x8 -1 --subme 5 --trellis 1 --level 3.0 --progress
simonhowson
18th April 2008, 08:40
Wow, this gets even more complicated.
I have TWO iPods, an 80 GB 5.5G, and a 160 GB Classic (I guess this counts as the 6th Generation).
Thank you to Ranguvar for recommending I encode the files anamorphic. The quality on the TV looks excellent, far better than the non-anamorphic encodes. In fact when sitting a few metres from the screen, the quality looks close to the DVDs.
Now I encoded 640 wide anamorphic files using the following command line:
--pass 2 --bitrate 1000 --stats ".stats" --level 3 --no-cabac --subme 6 --partitions none --qpmin 16 --vbv-bufsize 1000 --vbv-maxrate 2500 --me umh --merange 12 --threads auto --thread-input --progress --no-psnr --no-ssim --output "output" "input"
The files play perfectly on both iPods, however! The 5.5G iPod will not output the files to a TV. I wish I understood why! This isn't the end of the world, I bought the iPod Classic to replace the 5.5G.
The second point is, setting VBV max to 10000 causes problems on the 5.5G when the encode has deblocking. I appreciate Southstorm's comment that:
The point is that a Level 3.0 profile requires the device to handle a bitrate "spike" every now and then up to 10000
But it seems that the 5.5 has trouble simultaneously decoding high bitrate spikes, and running the deblocking filter.
It seems that the 5.5G processor just isn't fast enough to do all that work at once. I have encoded a clip that includes an aircraft dog fight, and the decoding pauses breifly at these points in the clip. It quickly resumes normal speed as the scene complexity declines, but I want to make sure my encodes play perfectly.
Thank you Southstorm for the suggested command line in the above post. I'll give that ago as well.
I'm trying to work out my encoding system because I want to back up my entire DVD collection.
Southstorm
18th April 2008, 11:46
Well, my wife has an 80GB iPod Classic, so I've been experimenting...
I encoded a 10 min sample of U-571 with lots of motion with this exact commandline:x264.exe --progress --no-psnr --no-cabac --cqm flat --no-ssim --no-b-adapt --scenecut 40
-I 250 -i 25 --me umh --merange 16 --sar 1:1 --threads auto -r 1 --nr 0 -f 0:0 -b 0
--direct "spatial" --b-bias 0 --direct-8x8 -1 -p 1 -B 1500 -A p8x8,i4x4,p4x4 -m 6
--vbv-init 0.9 --vbv-bufsize 10000 --vbv-maxrate 10000 --level 3.0 -o "output.264" "iPod.avs"
The deblock was at default setting 0:0, and it plays fine on the iPod.
@Simonhowson
Are you sure there aren't other settings in the commandline that is causing the stutter? Could you post the line?
simonhowson
19th April 2008, 03:18
Are you sure there aren't other settings in the commandline that is causing the stutter? Could you post the line?
My iPod CLASSIC can play the files fine. The iPod 5.5G seems to struggle with high bitrate and deblocking on.
I use the encoder settings in my post above.
Southstorm
21st April 2008, 13:34
Sorry, missed your earlier post...
I'd add "-b 0" for NO B frames, since a Baseline profile is required. And "-r 1" for 1 reference frame. See if this makes the stream more stable for you 5.5G iPod.
simonhowson
21st April 2008, 14:32
Sorry, missed your earlier post...
I'd add "-b 0" for NO B frames, since a Baseline profile is required. And "-r 1" for 1 reference frame. See if this makes the stream more stable for you 5.5G iPod.
Excellent, I'll give it a go. Thanks for your suggestion.
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.