HandBrake and Palm thread - Page 6

Closed Thread
Page 6 of 9 FirstFirst ... 23456789 LastLast
Results 51 to 60 of 84
  1. #51
    Brain stuck BogoMipping
    Join Date
    Feb 2004
    Location
    Medellin, Colombia
    Posts
    7,106

    Default Re: HandBrake and Palm thread

    Quote Originally Posted by FireNWater View Post
    Also, .avi options will be going away in v9.4 so anyone like me who needs to learn more H.264 and mp4 and overclocking before moving on needs to archive a 9.3 or 9.2 for backup.
    FNW, HB is dumping AVI, but this doesn't mean you will have to use H.264. You can still use XviD or FFMPEG in an MP4 container. With these two you can play high bitrate videos on Palm devices without the need to overclock.
    raspabalsa
    Palm Vx -> Palm m515 -> Palm Tungsten T1 -> Palm Tungsten T2 -> Palm Tungsten T3 -> Palm TX -> Samsung Galaxy Player 5.0 + Apple BT Keyboard -> Sony Xperia Z1 phone + Logitech BT Keyboard

    There are just four simple machines to alter force: the lever, the pulley, the inclined plane and, um, the internal combustion engine - Calvin
    Anything with a large enough engine will fly

  2. #52
    Mobile Enthusiast
    Join Date
    Sep 2008
    Location
    Bay Area, CA
    Posts
    108

    Default Re: HandBrake and Palm thread

    ""Other options (detelecine, decomb, deinterlace, denoise, deblock) were left disabled, since I'm not yet familiar with some of them."

    I was wrong about this stuff just being for broadcasting. If we are ripping to run on a PMP, then we want to turn off the lacing since we are not broadcasting. If the source content is interlaced, it needs to be de-interlaced so that it will look good on our PMPs.

    "HB info: Interlacing only looks right on an old tube TV (Cathode Ray Tube, or CRT).
    If you have an LCD or a plasma, interlaced content looks horrible. "

    http://trac.handbrake.fr/wiki/DeinterlacingGuide

    Also: HandBrake now offers a new filter, called Decomb, that can be used instead of Deinterlace.

    It lists how to tell if the source is interlaced and such.

    [rant] I can't help following this thread cuz it's way more interesting than reading for my Computer Op Sys and Hardware textbook. I'm annoyed that our hardcopy texts are 7 yrs out of date for online classes but every tech tells me the tests/core materials won't change that much for A+Certif but why bother learning out of use stuff like Win 9x/ME which are no longer supported by Win. ugghhhh. [/rant]

    Having been warned by Raz to avoid the forums which are bleak, just been digging thru the wicki instead.

    OK back to the grind after having a fun break on HB and experimenting with the specs from your projects (errr tests).
    LTS: life's too short.
    Palms: very old freebie, V, E2, T/X

  3. #53
    Mobile Enthusiast
    Join Date
    Sep 2008
    Location
    Bay Area, CA
    Posts
    108

    Default Re: HandBrake and Palm thread

    for those following this in future so they don't have to jump to another url.

    Deinterlace

    If you see combing or teeth (horizontal lines) in the preview images, choose a method from the Deinterlace pop-up menu. If that makes the lines go away, keep it enabled. For a fuller explanation, see the Deinterlacing Guide.

    Be aware that there is no difference in the previews no matter which deintrlacing method you choose. Single frame previews can only be filtered by "Fast." Even though you can't tell a difference from what happens to the previews, fast deinterlacing seriously reduces picture quality compared to the slow methods, while the slow deinterlacing methods will seriously reduce your encoding speeds compared to the fast method.

    If you don't see horizontal lines, and enabling deinterlacing doesn't seem to improve the picture in the preview frames, disable it. You're better off leaving it disabled except when it's absolutely needed.

    "Fast deinterlacing" is the FFmpeg project's internal deinterlacer, avpicture_deinterlace. "Slow" and "Slower" are yadif, from the MPlayer project.
    Decomb

    If you see combing or mouse teeth (horizontal lines) in the preview images, you should apply deinterlacing.

    But deinterlacing is slow and reduces picture quality. And a lot of the time it isn't necessary. And it's not always easy to know a video is interlaced just from looking at previews. And then you waste a bunch of time deinterlacing frames in interlaced video that don't even show combing, slowing down your encode.

    Wouldn't it be nice you could tell HandBrake to just deinterlace a frame if it's interlaced?

    That's what Decomb does. It should be safe to leave it on all the time.
    Last edited by FireNWater; 08-01-2009 at 10:01 PM. Reason: correct spelling errors
    LTS: life's too short.
    Palms: very old freebie, V, E2, T/X

  4. #54
    Brain stuck BogoMipping
    Join Date
    Feb 2004
    Location
    Medellin, Colombia
    Posts
    7,106

    Default Re: HandBrake and Palm thread

    Quote Originally Posted by FireNWater View Post
    I was wrong about this stuff just being for broadcasting. If we are ripping to run on a PMP, then we want to turn off the lacing since we are not broadcasting. If the source content is interlaced, it needs to be de-interlaced so that it will look good on our PMPs.

    "HB info: Interlacing only looks right on an old tube TV (Cathode Ray Tube, or CRT).
    If you have an LCD or a plasma, interlaced content looks horrible. "

    http://trac.handbrake.fr/wiki/DeinterlacingGuide

    Also: HandBrake now offers a new filter, called Decomb, that can be used instead of Deinterlace.
    Thanks, that's good info. I see it comes from HB's online guide. I'm downloading it now to see what else it has to say about the advanced options.
    Quote Originally Posted by FireNWater View Post
    [rant] I can't help following this thread cuz it's way more interesting than reading for my Computer Op Sys and Hardware textbook. I'm annoyed that our hardcopy texts are 7 yrs out of date for online classes but every tech tells me the tests/core materials won't change that much for A+Certif but why bother learning out of use stuff like Win 9x/ME which are no longer supported by Win. ugghhhh. [/rant]
    You're studying about ME? Heh heh... that qualifies as core material that won't change?
    Quote Originally Posted by FireNWater View Post
    Wouldn't it be nice you could tell HandBrake to just deinterlace a frame if it's interlaced?

    That's what Decomb does. It should be safe to leave it on all the time.
    Ah, that's interesting. I just went through that part of the guide. I'll have to check if the decomb filter is as good as advertised. I suppose enabling it will have a major impact on encoding time, will the result be worth it? And if my source video is not interlaced, why should I keep it on? More to the point, can I positively identify an interlaced video from a non-interlaced one, given that my monitor is a CRT and does not show the artifacts an interlaced video shows on a LCD?
    raspabalsa
    Palm Vx -> Palm m515 -> Palm Tungsten T1 -> Palm Tungsten T2 -> Palm Tungsten T3 -> Palm TX -> Samsung Galaxy Player 5.0 + Apple BT Keyboard -> Sony Xperia Z1 phone + Logitech BT Keyboard

    There are just four simple machines to alter force: the lever, the pulley, the inclined plane and, um, the internal combustion engine - Calvin
    Anything with a large enough engine will fly

  5. #55
    Brain stuck BogoMipping
    Join Date
    Feb 2004
    Location
    Medellin, Colombia
    Posts
    7,106

    Default Re: HandBrake and Palm thread

    Quote Originally Posted by Varjak View Post
    Frankly (and I'll have to play with this) I doubt that the overclocking makes that much of a difference (from essentially unplayable to smooth-playing). Maybe the newest HB is that much better. I don't know.
    Varjak, I found some very interesting info about this. Today I tested my H.264 videos on a friend's T5, and was very surprised by the results. His T5 runs at stock speed (416 MHz) and at this speed TCPMP was completely unable to play the H.264 videos that play smoothly on my TX at the same CPU clock. For example, the 600 kbps H.264 that benchmarked about 101% on my TX (at 416 MHz) scored an incredibly lousy 46% on the T5 at the same clock. Playback was next to impossible, with the image freezing, the audio falling out of sync, all the symptoms you've described on your LD. Then we increased his CPU clock to 520 MHz, and the benchmark improved nearly nothing at all. Playback still as bad as at 416 MHz. This was a real shock, since we expected the T5 to perform at least as well as the overclocked TX, and perhaps even better. I reinstalled TCPMP (the same vesion I used on my TX) and all the codecs, and things didn't improve. We ran several tests, including right after a soft reset, after a warm and soft reset, flushing cache, etc. The results were the same in every case.

    I then tested the videos on CorePlayer on his T5, and guess what? The H.264 videos that failed so miserably on TCPMP play perfectly well on CorePlayer. Benchmarks and playback tests at 416 MHz and 520 MHz yielded results very similar to the ones I got on my TX at those clocks. Playback was smooth, audio was perfectly synced, just like on my TX.

    I don't know what may cause this difference, but clearly something in the combination T5/TCPMP makes the H.264 videos play so bad, while CorePlayer for some reason can perfectly cope with them. I wonder if this same thing happens in the LD, given that yours shows the same symptoms as my friend's T5 when using TCPMP. Could it be that CorePlayer will play H.264 smoothly on your LD?

    This is very strange, especially since on my TX TCPMP performed better than CorePlayer, benchmarking about 10% more in every test. Why would the situation be so radically reversed on the T5, with TCPMP benchmarking almost 50% below CP? The TX and T5 are very similar in hardware, and the OS can't be that much different either. And what about the LD? It's the same CPU, similar hardware (except for the HDD and associated electronics), but can it be so much different from our devices?

    So the real difference seems to be not in the newer HB version, but in our own handhelds. If this results are accurate (and I have no way to further test because no other Palm around here uses CorePlayer) then it seems that if you want to play H.264 on a T5 (and possibly an LD, if verified) then your only option is to use CorePlayer. Too bad that CP is payware, with no trial period. You would have to purchase if you want to test, and there's no guarantee that it will work ok.
    Last edited by raspabalsa; 08-02-2009 at 12:30 AM. Reason: typos
    raspabalsa
    Palm Vx -> Palm m515 -> Palm Tungsten T1 -> Palm Tungsten T2 -> Palm Tungsten T3 -> Palm TX -> Samsung Galaxy Player 5.0 + Apple BT Keyboard -> Sony Xperia Z1 phone + Logitech BT Keyboard

    There are just four simple machines to alter force: the lever, the pulley, the inclined plane and, um, the internal combustion engine - Calvin
    Anything with a large enough engine will fly

  6. #56
    Brain stuck BogoMipping
    Join Date
    Feb 2004
    Location
    Medellin, Colombia
    Posts
    7,106

    Default Re: HandBrake and Palm thread

    Much as I don't like to post so many consecutive posts, I wanted to add that I have some more results from my H.264 tests. Last night I encoded my test video with H.264 with the "Anamorphic loose" option enabled. Videos encoded with this option produce a major performance hit on my TX. The original framerate of the video is 24 fps, but with the video encoded as anamorphic the TX can sustain only 15.5 fps even when the TX is overcloked to 520 MHz. This translates into very choppy playback, especially at high action or panning scenes, of which my test clip has many. The same clip encoded at the same bitrate and without enabling anamorphic plays very smoothly on my TX. I'm not yet familiar with how anamorphic works, but I suspect it adds aspect ratio information on a per-frame basis. Or maybe the implementation of the H.264 codec on TCPMP and CorePlayer force the TX to stretch each frame on the fly?

    I'm still not sure what's the advantage of encoding videos using this anamorphic option. Both videos are played on my TX at the proper aspect ratio, meaning there is no vertical or horizontal squishing.

    I've been encoding other videos following Varjak's advise for a fair trial. I have been lowering the video bitrate used by H.264, and results look promising. Even at bitrates as low as 200 kbps, H.264 video looks acceptable, with a quality similar to a 350 kbps XviD video. I'll post more precise results (including the full continuous movie playback Varjak suggested earlier) soon.
    raspabalsa
    Palm Vx -> Palm m515 -> Palm Tungsten T1 -> Palm Tungsten T2 -> Palm Tungsten T3 -> Palm TX -> Samsung Galaxy Player 5.0 + Apple BT Keyboard -> Sony Xperia Z1 phone + Logitech BT Keyboard

    There are just four simple machines to alter force: the lever, the pulley, the inclined plane and, um, the internal combustion engine - Calvin
    Anything with a large enough engine will fly

  7. #57
    Hookette's edgy lately
    Join Date
    Jul 2003
    Location
    Northern VA, USA
    Posts
    16,699

    Default Re: HandBrake and Palm thread

    Keep posting as many consecutive posts as you want. This is a great resource for my own Handbrake experiments (for both the TX and 5800). I had thought I wold get to them this weekend, but life intervened. I will try next week-end.
    Hook's Palm TX Help Page

    Past: Palm T2 > Palm TX > Nokia 5800 > HTC Nexus One > Samsung Galaxy Player 5 > Nexus 7 (2012) > LG Nexus 4 (gave to son) > LG Nexus 5

    Current:
    Nexus 5, 32 GB,Rooted, Dirty Unicorns ROM 7.5, Android 4.4.4, Franco kernel.
    Nexus 7 (2012) 3G (converted to Grouper wifi only), Rooted, Dirty Unicorns 7.5, Android 4.4.4, Franco Kernel.

  8. #58
    Banned
    Join Date
    Dec 2006
    Posts
    6,428

    Default Re: HandBrake and Palm thread

    Raspy, your 2 posts above lead me to conclude that the issue with playback is more a result of the anamorphic mode than the difference in devices. If I get a chance, I'll try encoding a non-anamorphic sample and see if that makes a difference. Also, is it possible that in your friend's case, the two TCPMP's were not set up the same way (options, etc.; not installed plugins).

    The reason different devices might behave differently is more than clock speed. I was told, for example, that the drivers are different between the TX and the LD with respect to Softick Audio Gateway. I noticed there was a footnote that sometimes the LD exhibited 'skipping' with SAG, while the TX is unaffected. I emailed them, surprised that two devices, so similar, would behave differently (assuming an SD card as the source for both). They told me they use different drivers. The T5 may be similar. I don't know what version of the OS the T5 uses; but maybe that affects the result as well.

    I'm a bit surprised that anamorphic would make that difference. The way I understand it; the anamorphic setting should be a shortcut of sorts; but maybe the calculations of formatting it into the appropriate dimensions is too intensive.

    What were the differences in dimensions between the two trials you did (anamorphic v. not anamorphic)? Do you get a properly formatted 'widescreen' picture? I'm not that anal that I need the result to be perfect; but I'd prefer to have the entire film image and 'fit it' to the horizontal dimension of 480. That only makes sense to me. The anamorphic section at HB confused me to be honest and when I tried to ask a question about it, I got shelled. I tried to find out if I could just encode the 'base' image; but that got me into a lot of trouble.
    Last edited by Varjak; 08-03-2009 at 05:22 PM.

  9. #59
    Banned
    Join Date
    Dec 2006
    Posts
    6,428

    Default Re: HandBrake and Palm thread

    I don't have the time to find that thread. It was from the guy who posts the long, in-depth articles about Bluetooth, browsers, etc. In the Wikipedia article, it seems that AVC has a bunch of 'steps' in its standard, and Palms only support the most basic AVC.

    The things you quote from HB weren't posted a year ago when I started with this. Obviously, they've upgraded this stuff a lot. They did not label the MPEG4 codec as ffmpeg (which evidently it is). I thought I remembered reading that XviD was inferior, which is why I went with MPEG4. It yields the best result for me by far.

    I'm not going to mess with overclocking. I know people here tend to add stuff and it all works swimmingly. I add a heavily vetted app and my whole LD goes to shyte.

  10. #60
    Banned
    Join Date
    Dec 2006
    Posts
    6,428

    Default Re: HandBrake and Palm thread

    Another thing. I thought others with TX's had tried AVC, so I didn't think minor device differences were a problem; but maybe they are. I know people have mentioned that the LD uses it's memory in a less efficient way, which might have an effect. I've read that TCPMP has buffering options, etc. that might make a difference. I wouldn't mind if I had to let the device buffer some before watching if it helped.

 

 

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  
All times are GMT -4. The time now is 11:43 PM.
Powered by vBulletin® Version 4.2.2
Copyright © 2014 vBulletin Solutions, Inc. All rights reserved.
Search Engine Friendly URLs by vBSEO 3.6.1