ATV Advanced Settings (HD Source)

Discuss encoding for devices and presets.
Forum rules
An Activity Log is required for support requests. Please read How-to get an activity log? for details on how and why this should be provided.
Cavalicious
Moderator
Posts: 1804
Joined: Mon Mar 26, 2007 12:07 am

ATV Advanced Settings (HD Source)

Post by Cavalicious »

OK, Here is the baseline to build off of. This has been tested, and works like a charm on the ATV.

Note: AppleTV only supports 30 fps video sources @ 960x540 resolution. While iTunes supports the native HD resolution.


Image

Syntax:

Code: Select all

./HandBrakeCLI -i /Volumes/Encode/TheTonightShow.mpg  -o /Volumes/New\ Downloads/HD_Video.m4v  -w 960 -l 540 -m  -F -q 0.65 -Q -e x264 -r 29.97   --deinterlace="1" -E facc  -B 384 -R 44.1 -6 6ch -v -x keyint=300:keyint-min=30:bframes=6:ref=3:mixed-refs=1:subq=5:me=umh:no-fast-pskip=1:trellis=0:no-dct-decimate=1:vbv-maxrate=4900:vbv-bufsize=2500
Log:

Code: Select all

[15:47:30] hb_init: checking cpu count
[15:47:30] hb_init: starting libhb thread
[15:47:30] thread 1804200 started ("libhb")
HandBrake 0.9.0 (2007081900) - http://handbrake.m0k.org/
2 CPUs detected
Opening /Volumes/Encode/TheTonightShow.mpg...
[15:47:30] hb_scan: path=/Volumes/Encode/TheTonightShow.mpg, title_index=1
[15:47:30] thread 1804600 started ("scan")
[15:47:30] scan: trying to open with libdvdread
ERROR: dvd: DVDOpen failed (/Volumes/Encode/TheTonightShow.mpg)[15:47:30] scan: trying to open as MPEG-2 Stream
[15:47:30] scan: decoding previews for title 0
[15:47:30] scan: preview 1
[15:47:30] scan: AC3, rate=48000Hz, bitrate=384000
[15:47:30] scan: preview 2
[15:47:30] scan: preview 3
[15:47:30] scan: preview 4
Scanning title 0...
[15:47:30] scan: preview 5
[15:47:30] scan: preview 6
[15:47:30] scan: preview 7
[15:47:31] scan: preview 8
Scanning title 0...
[15:47:31] scan: preview 9
[15:47:31] scan: preview 10
[15:47:31] scan: 1920x1088, 29.970 fps, autocrop = 0/0/0/0
[15:47:31] hb_stream_update_audio: id=80bd, lang=Unknown (AC3) (5.1 ch), 3cc=und, rate = 48000, bitrate = 384000, flags = 0x17 (23)
[15:47:31] scan: title (0) job->width:1920, job->height:1088
[15:47:31] thread 1804600 exited ("scan")
[15:47:31] thread 1804600 joined ("scan")
[15:47:31] libhb: scan thread found 1 valid title(s)
+ title 0:
  + vts 0, ttn 0, cells 0->0 (0 blocks)
  + duration: 00:00:49
  + size: 1920x1088, aspect: 1.78, 29.970 fps
  + autocrop: 0/0/0/0
  + chapters:
    + 1: cells 0->0, 0 blocks, duration 00:00:49
  + audio tracks:
    + 1, Unknown (AC3) (5.1 ch), 48000Hz, 384000bps
  + subtitle tracks:
Applying the following x264 options: keyint=300:keyint-min=30:bframes=6:ref=3:mixed-refs=1:subq=5:me=umh:no-fast-pskip=1:trellis=0:no-dct-decimate=1:vbv-maxrate=4900:vbv-bufsize=2500
[15:47:31] thread 1804600 started ("work")
[15:47:31] 1 job(s) to process
[15:47:31] starting job
[15:47:31]  + device /Volumes/Encode/TheTonightShow.mpg
[15:47:31]  + title 0, chapter(s) 1 to 1
[15:47:31]  + 1920x1088 -> 960x540, crop 0/0/0/0
[15:47:31]  + grayscale off
[15:47:31]  + filters
[15:47:31]    + Deinterlace (ffmpeg or yadif/mcdeint) (1)
[15:47:31]  + 29.970 fps, video quality 0.65
[15:47:31]  + PixelRatio: 0, width:960, height: 540
[15:47:31]  + encoder x264
[15:47:31]  + audio 384 kbps, 44100 Hz
[15:47:31]  + encoder faac
[15:47:31]    + 80bd, Unknown (AC3) (5.1 ch)
[15:47:31]      + Requested mixdown: 6-channel discrete (HB_AMIXDOWN_6CH)
[15:47:31]      + Actual mixdown: 6-channel discrete (HB_AMIXDOWN_6CH)
[15:47:31] thread 1806a00 started ("reader")
[15:47:31]  + output: /Volumes/New Downloads/HD_Video.m4v
ERROR: dvd: DVDOpen failed (/Volumes/Encode/TheTonightShow.mpg)[15:47:31] thread 1807400 started ("muxer")
[15:47:31] thread 1816a00 started ("MPEG-2 decoder (libmpeg2)")
[15:47:31] thread 1c5d200 started ("Renderer")
[15:47:31] encx264: Encoding at constant RF 17.850002
[15:47:31] encx264: opening libx264 (pass 0)
x264 [warning]: width or height not divisible by 16 (960x540), compression will suffer.
x264 [info]: using cpu capabilities: MMX MMXEXT SSE SSE2 
[15:47:31] thread 1ca9e00 started ("H.264/AVC encoder (libx264)")
No accelerated IMDCT transform found
[15:47:31] thread 1ca9a00 started ("AC3 decoder")
[15:47:31] thread 1cbc800 started ("AAC encoder (libfaac)")
[15:47:31] sync: expecting 1499 video frames
[15:47:31] sync: first pts is 13206
Encoding: task 1 of 1, 92.19 % (4.23 fps, avg 4.38 fps, ETA 00h00m26s)[15:52:47] reader: done
[15:52:47] thread 1806a00 exited ("reader")
Encoding: task 1 of 1, 98.67 % (4.50 fps, avg 4.38 fps, ETA 00h00m05s)[15:53:09] sync: got 1480 frames, 1499 expected
Encoding: task 1 of 1, 98.73 % (4.50 fps, avg 4.38 fps, ETA 00h00m05s)[15:53:16] thread 1ca9e00 exited ("H.264/AVC encoder (libx264)")
[15:53:16] thread 1ca9a00 exited ("AC3 decoder")
[15:53:16] thread 1816a00 exited ("MPEG-2 decoder (libmpeg2)")
[15:53:16] thread 1cbc800 exited ("AAC encoder (libfaac)")
[15:53:16] thread 1816a00 joined ("MPEG-2 decoder (libmpeg2)")
[15:53:16] mux: file size, 18758853 bytes
[15:53:16] mux: track 0, 17542918 bytes, 2859.34 kbps
[15:53:16] mux: track 1, 1176680 bytes, 191.79 kbps
[15:53:16] mux: overhead, 10.96 bytes per frame
[15:53:16] thread 1807400 exited ("muxer")
[15:53:16] thread 1c5d200 exited ("Renderer")
[15:53:16] thread 1c5d200 joined ("Renderer")
[15:53:16] fifo_close: trashing 1 buffer(s)
[15:53:16] thread 1ca9e00 joined ("H.264/AVC encoder (libx264)")
x264 [info]: slice I:8     Avg QP:18.12  size: 96039  PSNR Mean Y:45.01 U:47.50 V:47.27 Avg:45.66 Global:45.42
x264 [info]: slice P:436   Avg QP:19.27  size: 25043  PSNR Mean Y:43.99 U:46.80 V:46.44 Avg:44.70 Global:44.62
x264 [info]: slice B:1027  Avg QP:21.40  size:  5701  PSNR Mean Y:43.30 U:46.70 V:46.33 Avg:44.13 Global:44.06
x264 [info]: mb I  I16..4: 14.8%  0.0% 85.2%
x264 [info]: mb P  I16..4:  2.5%  0.0%  3.0%  P16..4: 56.2% 25.0% 10.7%  0.0%  0.0%    skip: 2.5%
x264 [info]: mb B  I16..4:  0.1%  0.0%  0.5%  B16..8: 17.0%  2.0%  4.3%  direct: 8.5%  skip:67.6%
x264 [info]: ref P  79.1% 11.7%  9.1%
x264 [info]: ref B  88.6%  7.2%  4.2%
x264 [info]: SSIM Mean Y:0.9811665
x264 [info]: PSNR Mean Y:43.514 U:46.734 V:46.365 Avg:44.306 Global:44.226 kb/s:2859.25
[15:53:16] thread 1ca9a00 joined ("AC3 decoder")
[15:53:16] thread 1cbc800 joined ("AAC encoder (libfaac)")
[15:53:16] thread 1806a00 joined ("reader")
[15:53:16] thread 1807400 joined ("muxer")
[15:53:16] fifo_close: trashing 0 buffer(s)
[15:53:16] fifo_close: trashing 0 buffer(s)
[15:53:16] fifo_close: trashing 0 buffer(s)
[15:53:16] fifo_close: trashing 1 buffer(s)
[15:53:16] fifo_close: trashing 0 buffer(s)
[15:53:16] Freed 4 buffers of size 512
[15:53:16] Freed 2 buffers of size 1024
[15:53:16] Freed 2048 buffers of size 2048
[15:53:16] Freed 0 buffers of size 4096
[15:53:16] Freed 13 buffers of size 8192
Muxing: 0.00 %[15:53:16] Freed 0 buffers of size 16384
[15:53:16] Freed 0 buffers of size 32768
[15:53:17] Freed 137 buffers of size 3133440
[15:53:17] Allocated 433750016 bytes of buffers on this pass and Freed 433586176 bytes, 163840 bytes leaked
[15:53:17] fifo_close: trashing 0 buffer(s)
[15:53:17] fifo_close: trashing 0 buffer(s)
[15:53:17] fifo_close: trashing 0 buffer(s)
[15:53:17] fifo_close: trashing 20 buffer(s)
[15:53:17] thread 1804600 exited ("work")
[15:53:17] thread 1804600 joined ("work")
[15:53:17] libhb: work result = 0

Rip done!
[15:53:17] thread 1804200 exited ("libhb")
[15:53:17] thread 1804200 joined ("libhb")
HandBrake has exited.
Specs:
  • Original MPEG Program Stream = 94MB (50secs)
    Original MPEG Program Stream bitrate = 15.94 mbits/s
    Converted .264 m4v = 17.9MB
    Converted .264 bitrate (Video) = 2854.82 kbits/s
    Converted .264 bitrate (Audio) = 192.18 kbits/s
Cavalicious
Moderator
Posts: 1804
Joined: Mon Mar 26, 2007 12:07 am

Post by Cavalicious »

I'm currently encoding a complete program (AST Dew Tour - Toyota Challenge). I suspect a lot of panning and granularity, so I'll see if the same settings bend the ATV or not.

@ 1hr30min source and averaging 4.17fps... may have to post results tomorrow. 10 more hours to go!
Cavalicious
Moderator
Posts: 1804
Joined: Mon Mar 26, 2007 12:07 am

Post by Cavalicious »

Source :

AST Dew Tour - Toyota Challenge (MPEG Program Stream)
1920 x 1080 pixels
10.12 GB
15.78 mbits/s

.264 Encode:

960 x 540 pixels
2.87 GB
Video bitrate - 4208.46 kbits/s
Audio bitrate - 268.54 kbits/s

I got a mess load of these during the encode:

Code: Select all

Encoding: task 1 of 1, 16.31 % (4.47 fps, avg 4.15 fps, ETA 09h15m04s)[18:10:42] 899.628601s: Progressive -> Interlaced
Encoding: task 1 of 1, 16.31 % (4.49 fps, avg 4.15 fps, ETA 09h15m03s)[18:10:43] 899.762146s: Interlaced -> Progressive
Encoding: task 1 of 1, 16.36 % (4.36 fps, avg 4.15 fps, ETA 09h14m38s)[18:11:00] 902.331360s: Progressive -> Interlaced
Encoding: task 1 of 1, 16.36 % (4.46 fps, avg 4.15 fps, ETA 09h14m35s)[18:11:02] 902.631653s: Interlaced -> Progressive
Encoding: task 1 of 1, 16.40 % (4.31 fps, avg 4.15 fps, ETA 09h14m14s)[18:11:16] 904.750427s: Progressive -> Interlaced
Encoding: task 1 of 1, 16.40 % (4.29 fps, avg 4.15 fps, ETA 09h14m13s)[18:11:17] 904.883911s: Interlaced -> Progressive
Encoding: task 1 of 1, 16.47 % (4.40 fps, avg 4.15 fps, ETA 09h13m37s)[18:11:41] 908.637695s: Progressive -> Interlaced
Encoding: task 1 of 1, 16.48 % (4.48 fps, avg 4.15 fps, ETA 09h13m32s)[18:11:45] 909.004700s: Interlaced -> Progressive
Encoding: task 1 of 1, 16.60 % (4.48 fps, avg 4.15 fps, ETA 09h12m24s)[18:12:29] 915.694641s: Progressive -> Interlaced
Encoding: task 1 of 1, 16.60 % (4.58 fps, avg 4.15 fps, ETA 09h12m22s)[18:12:30] 915.794739s: Interlaced -> Progressive
Encoding: task 1 of 1, 16.61 % (4.47 fps, avg 4.15 fps, ETA 09h12m19s)[18:12:32] 916.195190s: Progressive -> Interlaced
Encoding: task 1 of 1, 16.61 % (4.40 fps, avg 4.15 fps, ETA 09h12m18s)[18:12:33] 916.295288s: Interlaced -> Progressive
Encoding: task 1 of 1, 16.65 % (4.42 fps, avg 4.15 fps, ETA 09h12m00s)[18:12:45] 918.147095s: Progressive -> Interlaced
Encoding: task 1 of 1, 16.65 % (4.42 fps, avg 4.15 fps, ETA 09h12m00s)[18:12:46] 918.213867s: Interlaced -> Progressive
Encoding: task 1 of 1, 16.65 % (4.35 fps, avg 4.15 fps, ETA 09h11m57s)[18:12:48] 918.430725s: Progressive -> Interlaced
Encoding: task 1 of 1, 16.67 % (4.47 fps, avg 4.15 fps, ETA 09h11m47s)[18:12:55] 919.381714s: Interlaced -> Progressive
Encoding: task 1 of 1, 16.72 % (4.53 fps, avg 4.16 fps, ETA 09h11m20s)[18:13:11] 921.934204s: Progressive -> Interlaced
Encoding: task 1 of 1, 16.73 % (4.50 fps, avg 4.16 fps, ETA 09h11m11s)[18:13:17] 922.818481s: Interlaced -> Progressive
Encoding: task 1 of 1, 16.74 % (4.48 fps, avg 4.16 fps, ETA 09h11m10s)[18:13:19] 923.035400s: Progressive -> Interlaced
Encoding: task 1 of 1, 16.74 % (4.56 fps, avg 4.16 fps, ETA 09h11m06s)[18:13:21] 923.419128s: Interlaced -> Progressive
Encoding: task 1 of 1, 16.75 % (4.54 fps, avg 4.16 fps, ETA 09h11m01s)[18:13:25] 923.986328s: Progressive -> Interlaced
Encoding: task 1 of 1, 16.76 % (4.50 fps, avg 4.16 fps, ETA 09h10m56s)[18:13:27] 924.386780s: Interlaced -> Progressive
Encoding: task 1 of 1, 16.77 % (4.43 fps, avg 4.16 fps, ETA 09h10m51s)[18:13:31] 924.937256s: Progressive -> Interlaced
Encoding: task 1 of 1, 16.77 % (4.42 fps, avg 4.16 fps, ETA 09h10m50s)[18:13:32] 925.070740s: Interlaced -> Progressive
Encoding: task 1 of 1, 16.78 % (4.33 fps, avg 4.16 fps, ETA 09h10m48s)[18:13:33] 925.304260s: Progressive -> Interlaced
Encoding: task 1 of 1, 16.78 % (4.41 fps, avg 4.16 fps, ETA 09h10m45s)[18:13:35] 925.571228s: Interlaced -> Progressive
Encoding: task 1 of 1, 16.79 % (4.47 fps, avg 4.16 fps, ETA 09h10m42s)[18:13:38] 925.971619s: Progressive -> Interlaced
Encoding: task 1 of 1, 16.79 % (4.45 fps, avg 4.16 fps, ETA 09h10m38s)[18:13:40] 926.255310s: Interlaced -> Progressive
Encoding: task 1 of 1, 16.81 % (4.55 fps, avg 4.16 fps, ETA 09h10m32s)[18:13:44] 926.972595s: Progressive -> Interlaced
Encoding: task 1 of 1, 16.81 % (4.51 fps, avg 4.16 fps, ETA 09h10m28s)[18:13:46] 927.239563s: Interlaced -> Progressive
Encoding: task 1 of 1, 16.82 % (4.45 fps, avg 4.16 fps, ETA 09h10m25s)[18:13:49] 927.623291s: Progressive -> Interlaced
Encoding: task 1 of 1, 16.82 % (4.72 fps, avg 4.16 fps, ETA 09h10m21s)[18:13:51] 927.923584s: Interlaced -> Progressive
I suspect that its the transitions to/between/from commercials...

There was one major issue, and that was sound. It only came from the Left Speaker. Although Quicktime properties said in had all 6 channels. :?:

Image

Video detail was still stunning


Image

Scenes like these in the daytime cause very little if any jitter with the AppleTV. A neofight might not even notice.


Image

Similar scene (at night) produced no jitter what-so-ever.
Cavalicious
Moderator
Posts: 1804
Joined: Mon Mar 26, 2007 12:07 am

Post by Cavalicious »

@ dynaflash

I stated before that vbv-maxrate 4900 was giving me sync issues, well come to find out it was my source. (Having issues getting a good 1080i recording of Heroes) Did get a good 1080i recording of Pushing Daisies and used both 3900 and 4500 with a 2500 buffer. Both played exactly the same, a very slight hint of jitter on a couple of scenes. The bitrate of the MPEG PS was 15.8 mbits/s.

Since then, I have been using 4500 and haven't been having any issues (knock on wood).
dynaflash
Veteran User
Posts: 3820
Joined: Thu Nov 02, 2006 8:19 pm

Post by dynaflash »

Nice Cav, I have been using both, and they seem to work pretty well. I havent had any issues since using vbv settings with atv and crf. Very nice.
MichaelLAX

Post by MichaelLAX »

What are your source streams? DirecTiVO HR10-250 files? Firewire from cable STB? HD-DVD/Blu-Ray rips?
Cavalicious
Moderator
Posts: 1804
Joined: Mon Mar 26, 2007 12:07 am

Post by Cavalicious »

MichaelLAX wrote:What are your source streams? DirecTiVO HR10-250 files? Firewire from cable STB? HD-DVD/Blu-Ray rips?
Over the Air HD through EyeTV Hybrid. Specs are stated above.
Cavalicious
Moderator
Posts: 1804
Joined: Mon Mar 26, 2007 12:07 am

Post by Cavalicious »

@dynaflash

I was able to smooth the video out using these setting:

vbv-maxrate=4500:vbv-bufsize=3000

I used Monday's recording of Heroes (clee ;-)) with these settings and the video was smooth as butter. I did have an issue with the audio sync (slowly fell out of sync over time), but this was easily fixed by a quick pause and play. So the problem wasn't in the encode.

Could it be that the bufsize is too big? Then why is the video smooth? Its not like Audio and Video is streamed separately, then I could understand. But since the file is Muxed, I wouldn't suspect the buffer size...at least not at first.

When watching the encode in VLC, the bitrate never spikes above 6mb/s.
MichaelLAX

Post by MichaelLAX »

Cavalicious wrote:
MichaelLAX wrote:What are your source streams? DirecTiVO HR10-250 files? Firewire from cable STB? HD-DVD/Blu-Ray rips?
Over the Air HD through EyeTV Hybrid. Specs are stated above.
Great work! I hope to implement it soon...

Are you using Two-Pass; as I don't see the code for it listed? If not, why not; doesn't it always improve quality, at the same bit rate?

Also, I digitize non-OTA HD from my DirecTV DVRs using the ADS Tech box that results in 720x480 (anamorphic) Mpeg-2 files. Will this formula work in transoding these files, too (but of course not in true HD, but great quality 853x480))? Thanks

PS: Check out:
http://cgi.ebay.com/ws/eBayISAPI.dll?Vi ... 0171187221

I just purchased one, and I am waiting for it to arrive and install...
Cavalicious
Moderator
Posts: 1804
Joined: Mon Mar 26, 2007 12:07 am

Post by Cavalicious »

MichaelLAX wrote:Are you using Two-Pass; as I don't see the code for it listed? If not, why not; doesn't it always improve quality, at the same bit rate?
I'm using CRF not ABR
MichaelLAX wrote:Also, I digitize non-OTA HD from my DirecTV DVRs using the ADS Tech box that results in 720x480 (anamorphic) Mpeg-2 files. Will this formula work in transoding these files, too (but of course not in true HD, but great quality 853x480))
If you can get the files to your computer in a supported HD format...I sure it will.
MichaelLAX

Post by MichaelLAX »

Cavalicious wrote:If you can get the files to your computer in a supported HD format...I sure it will.
They come out as MPEG-2 720x480 (anamorphic to 853x480) as QuickTime .mpg files when imported. Does that qualify?
Cavalicious
Moderator
Posts: 1804
Joined: Mon Mar 26, 2007 12:07 am

Post by Cavalicious »

If its a PS, then yes.

Read this:

http://handbrake.m0k.org/forum/viewtopic.php?t=864
MichaelLAX

Post by MichaelLAX »

Cavalicious wrote:If its a PS, then yes.

Read this:

http://handbrake.m0k.org/forum/viewtopic.php?t=864
How do I determine if it is a Program Stream? It is just the 720x480 MPEG-2 output from the ADS Tech Instant DVD (for Mac) after digitizing. Currently I use VisualHub to convert to H.264 853x480 for use on my Apple TV.

Thanks for the link; i will study it soon...
MichaelLAX

Post by MichaelLAX »

OK, I read through your cited link and read up on the difference between MPEG program streams and transport streams on wikipedia, so all I need to know is how to change your line interface options to meet my needs.

I can see how to change the 960x540 to 848x480 (848 to keep it evenly divisible by 16). How do I change the 6 channel to 2 channel? Any other changes I would need to know?

Thanks.
Cavalicious
Moderator
Posts: 1804
Joined: Mon Mar 26, 2007 12:07 am

Post by Cavalicious »

./HandBrakeCLI -h
Anamonde
Regular User
Posts: 71
Joined: Fri Mar 16, 2007 11:11 pm

Post by Anamonde »

[quote="MichaelLAX"]
PS: Check out:
<Edited due to spamming>
MichaelLAX

Post by MichaelLAX »

[quote="Anamonde"][quote="MichaelLAX"]
PS: Check out:
<Edited due to Spamming>

Thank you.

I am well aware of that website. "Easy" for one, may not be so "easy" for another...

When I was younger, I was the first one on the block to update my OS, firmware, etc. I am well past the point of proving anything to myself, especially given the balance between my time if it does not work.

I make a good living and at this stage of life, I am happy to pay for tools that make my life easier.

"Free" is a relative concept; especially if you knew how much I value my time.

Einstein said that he never memorized anything that he could look up, because it took up too much valuable brain space. I began to understand what he meant when I turned 50.

Yes, I could study that website and do it myself, but I am saving brain space for other endeavors (like learning how to use HandBrakeCLI)...
Leo
Bright Spark User
Posts: 174
Joined: Thu Feb 22, 2007 4:39 pm

Post by Leo »

I like the Einstein quote.


Cavalicious:
subq=5:me=umh:no-fast-pskip=1:trellis=0:no-dct-decimate=1:
Just wondering, do you find benefit in subq=5 over subq=4?
I have read (and seemed to find in my own tests) that there is very little quality difference between the two, but subq=5 is noticable slower (at least 10% if my memory serves me correctly).

Also, do you think that no-fast-pskip=1 and no-dct-decimate=1 are worth it when you could maybe use subq=6 (enabling fancy bframe benefits) (or trellis=1) and use a slightly higher CRF?

(not disagreeing, just wondering!) :)

Also, when I am encoding from DVDs for AppleTV compatibility, should I also set the maxrate to 4500 and the bufsize to 3000? Do you notice any more blocking in any parts of your high-def encodes? I have tried using vbv limits with CRF for iPod encodes but ran into big problems on certain scene changes...
[http://handbrake.m0k.org/forum/viewtopic.php?t=3330]

Thanks
Cavalicious
Moderator
Posts: 1804
Joined: Mon Mar 26, 2007 12:07 am

Post by Cavalicious »

subq=5 -is just the default
subq=6 -where subq=5 is 35% slower that subq=1, subq=6 is 60%

for me, subq=5 is a good partner with ref=3

no-fast-pskip=1 and no-dct-decimate=1 -I use because I HATE blocks in my encodes.

I don't use Trellis, because of what could happen. Some say you shouldn't ever use trellis with CRF, and some say trellis=1 is safe. Really its up to you, I just never seen a need for it since I've been doing CRF.

For HD Sources I use this: :vbv-maxrate=4500:vbv-bufsize=3000

For DVD Sources I use this: :vbv-maxrate=4900:vbv-bufsize=3000

Both encodes run smooth as butter on AppleTV. I'm thinking about uping the bufsize a bit later on, to see if I can find a break point.
dynaflash
Veteran User
Posts: 3820
Joined: Thu Nov 02, 2006 8:19 pm

Post by dynaflash »

using 4500/3000 I did get a very minor stutter on Night at The Museum right when that first logo comes in in the intro. Not bad though.

I am always using crf 70%, if the buffering works there, then I know it will work with lower %.

Night works good as a test case as the stuttering issues show up right in the first five minutes. During the opening credits and the slow complex pans.
Leo
Bright Spark User
Posts: 174
Joined: Thu Feb 22, 2007 4:39 pm

Post by Leo »

dynaflash, is Night at The Museum HD or DVD?
Do you notice a significant difference between CRF 65% and 70%? Am I right in thinking it's about 50% bigger?

Cav:
Sorry to ask a Q that's below the level of this thread, but what is it that "could happen" with trellis in CRF? (I have heard mixed views but no explanation.)

And is there anything special about reference frames with subme=5 vs subme=4? (I ask because I almost always use subme=4 and would want to know if I'm missing something important!)

So you notice a difference in blockiness when adding no fast p skip and no dct decimate?

[sorry the next two paragraphs are a bit tangential]
Also, as I asked before: Do you notice any more blocking in any parts of your high-def encodes [when setting your vbv limits, compared to no vbv limits] ? I have tried using vbv limits with CRF for iPod encodes but ran into big problems on certain scene changes...
[http://handbrake.m0k.org/forum/viewtopic.php?t=3330]

Do you have an explanation for why the ipod limits cripple scene changes so much on the ipod when using CRF? I know the limits (maxrate 1500 and bufsize 2000) are a lot smaller, but so is the resolution (and framerate since I'm using PAL). It just seems odd if it's not an issue at all here..?

Thanks
Anamonde
Regular User
Posts: 71
Joined: Fri Mar 16, 2007 11:11 pm

Post by Anamonde »

MichaelLAX wrote:
Anamonde wrote:
MichaelLAX wrote: PS: Check out:
<Edited due to Spamming>

Thank you.

I am well aware of that website. "Easy" for one, may not be so "easy" for another...

When I was younger, I was the first one on the block to update my OS, firmware, etc. I am well past the point of proving anything to myself, especially given the balance between my time if it does not work.

I make a good living and at this stage of life, I am happy to pay for tools that make my life easier.

"Free" is a relative concept; especially if you knew how much I value my time.

Einstein said that he never memorized anything that he could look up, because it took up too much valuable brain space. I began to understand what he meant when I turned 50.

Yes, I could study that website and do it myself, but I am saving brain space for other endeavors (like learning how to use HandBrakeCLI)...
Sorry MichaelLAX, I was just trying to be helpful.
dynaflash
Veteran User
Posts: 3820
Joined: Thu Nov 02, 2006 8:19 pm

Post by dynaflash »

Leo wrote:dynaflash, is Night at The Museum HD or DVD?
Do you notice a significant difference between CRF 65% and 70%? Am I right in thinking it's about 50% bigger?
DVD. Actually regarding vbv buffer settings, x264 really could care less whether the source is dvd or HD afaik. Its the complexity and motion of the source.
Leo wrote:Do you notice a significant difference between CRF 65% and 70%? Am I right in thinking it's about 50% bigger?
Well, for my application and eyes, yes. But here again we go down the "video quality is subjective" rathole. I don't want to go down that rathole as its discussed way too much already. Having said that, 70% crf is my choice for testing vbv buffer limits mostly because it is likely as high as any reasonable person would encode at (okay, maybe some would argue 72%). And if the vbv limits work at 70% we know they will be effective at any lower setting. This is the gist of my vbv buffer quest for the atv.
dynaflash
Veteran User
Posts: 3820
Joined: Thu Nov 02, 2006 8:19 pm

Post by dynaflash »

Cavalicious wrote: For HD Sources I use this: :vbv-maxrate=4500:vbv-bufsize=3000

For DVD Sources I use this: :vbv-maxrate=4900:vbv-bufsize=3000

Both encodes run smooth as butter on AppleTV. I'm thinking about uping the bufsize a bit later on, to see if I can find a break point.
Cav, are you using a known "stuttering" source without vbv limits ? I am just curious as some sources of course will not stutter at all even without vbv buffering, and for 4900/3000 you allow for a 10900 kbps peak for a .5 sec duration, which is getting pretty close to the 12mb limit for the atv. Especially with cabac on.

Just curious how those settings have panned out.
Cavalicious
Moderator
Posts: 1804
Joined: Mon Mar 26, 2007 12:07 am

Post by Cavalicious »

dynaflash wrote: Cav, are you using a known "stuttering" source without vbv limits ? I am just curious as some sources of course will not stutter at all even without vbv buffering, and for 4900/3000 you allow for a 10900 kbps peak for a .5 sec duration, which is getting pretty close to the 12mb limit for the atv. Especially with cabac on.

Just curious how those settings have panned out.
So far so good...I plan on re-encoding Training Day and Last Samurai tonight (time permitting), since I can't find my other 2 problem children.
Locked