Daodan
3rd April 2006, 12:44
With latest megui+x264 I can't seem to select lossless or q=0. I think with an older version this was possible. I also looked at Sharktooth's losless profile, but it doesn't look that lossless to me, q=18, 3 b-frames, etc. I want something for transcoding lossless and huffyuv and also Lagarith don't want to work multithreaded in VD under any circumstances (even if Lagarith has a multithread option).
So what method can you suggest me, so I can get some speed out of my my dual cpu?
emmel
3rd April 2006, 13:03
Trivial way: Split your clip a into two parts b+c. Then, instead of
>x264 --threads 2 .... a
do
>start /belownormal /b x264 --threads 1 ... b
>start /belownormal /b x264 --threads 1 ... c
and join the results. The method works with every codec, not only x264, and scales absolutely linearly :)
Daodan
3rd April 2006, 13:08
Thanks for the tip. I'll keep that in mind, maybe I'll need it someday.
But you missed the point. I don't have problems with x264 multithread (I have doubts of how to use it as lossless), but with Lagarith and huffyuv in VD.
Daodan
3rd April 2006, 14:50
Sorry, I'm an idiot. It seems that I can choose lossless only if I choose High profile. Problem solved.
foxyshadis
3rd April 2006, 15:14
x264 lossless may be multithreaded, but even running at double speed it's still probably twice as slow or more, plus it decodes slower than lagarith (which is already slower than huffyuv). Of course you can probably fiddle the options to make it almost as fast as xvid if you sacrifice space.
The best lossless format I've found for my purposes is xvid with ehr matrix @q2 (plus you get working smp) and high-speed settings, as long as you can accept that some pixels might be +-1 rather than 100% identical. And it works in avi. Of course, I won't keep you from testing stuff.
Daodan
3rd April 2006, 15:41
Yes, you were right about that speed (no matter what settings, choosing q=0 kills any speed). Still huffy for me atm. Lagarith seems to compress really well, in YV12 mode at least, but it encodes too slow. I'll try your xvid + cqm method, 4 days for saving lossless is just too much.
ChronoCross
3rd April 2006, 17:51
The best lossless format I've found for my purposes is xvid with ehr matrix @q2 (plus you get working smp) and high-speed settings, as long as you can accept that some pixels might be +-1 rather than 100% identical. And it works in avi. Of course, I won't keep you from testing stuff.
The it's not lossless is it lol? With sharktooths profile you should be able to run th elossless encoder in realtime. at least from my testing it's like that. Akupenguin and him worked out some speed and quality issues some time back.
foxyshadis
3rd April 2006, 22:03
It's close enough that the accumulated error after 3-4 iterations is invisible comparing side-by-side and interleaved, psnr is in the mid or high 90s, and it has no effect at all on the final encode. It isn't perfectly lossless, but functionally lossless. Of course, the file size is 1/2-2/3 the size of lagarith, not 1/10, a dvd-res movie might still need 5-10 gigs.
The one caveat is that it's YV12 only, but so is x264.
Sharktooth
4th April 2006, 15:03
Oh, finally someone discovered EQM V3EHR ;)
bratao
4th April 2006, 17:47
I read that for loseless you should disable CABAC , Cabac at this insane bitrate kill the processator
Sharktooth
5th April 2006, 11:48
Disabling CABAC = loosing a lot of compression.
Audionut
5th April 2006, 15:49
Oh, finally someone discovered EQM V3EHR ;)
I always mistake it for Fox home matrix.;)
Sharktooth
5th April 2006, 20:18
yeah they're very similar but if you test psnr or SSIM, EHR has an edge... :)
however it's not correct to call it lossless but it so close to the source you have to amplify the difference between the source and the encoded frames to spot the actual difference.
Audionut
5th April 2006, 23:48
IIRC it was originally a cut and paste of the fox home matrix. Then you modified it with all 8's in the inter-frame.
IMHO, it would be better to modify the intra-frames.
Sharktooth
6th April 2006, 13:40
i couldnt find better intra coefficients distribution than that
Audionut
6th April 2006, 23:06
Modified matrix from Star Wars Ep3.
Which in itself, is modified fox home.
8,5,7,9,11,13,14,17
5,6,8,11,13,14,17,18
7,9,10,11,16,17,17,16
9,10,13,13,14,17,18,20
10,13,13,14,16,17,20,24
13,13,14,16,17,20,24,29
13,13,14,17,19,23,28,34
13,14,17,19,23,28,34,41
8,5,7,9,10,10,11,11
5,6,9,10,10,11,11,12
7,9,10,10,11,11,12,12
9,10,10,11,11,12,13,13
10,10,11,11,12,13,13,13
10,11,11,12,13,13,14,14
11,11,12,13,13,14,14,14
11,12,12,13,13,14,14,14
Sharktooth
7th April 2006, 03:07
AFAIK values below 8 can only cause problems in MPEG4 ASP and may even be not allowed...
Audionut
7th April 2006, 09:24
Yes they are allowed.
Only "problem" as such is high bitrate.:)
vBulletin® v3.8.5, Copyright ©2000-2012, Jelsoft Enterprises Ltd.