How to speed up Xvid encoding?

[Gladiator:)]

Hi another suggestion ? can u ? :helpful:

I do my rips in some quality plugins like LImitedSharpenFaster , SeeSaw etc
I know they take a much of time .
But it is really slow , takes 10 - 14 hrs sometimes 18
The first pass average Fr/sec is 23
In second pass it is 5-7

I have a Xp 2.66 GHz and a 759 MB ram speed
While ripping i do not do any work and i don't care abt stabilty of system .
Is there any way to increase the speed.

Don't take it too technical pl:p

[sh0dan]

:script:

.. but it sounds like the "problem" is the encoder, and not primarily your script. So you might be better off asking in the proper encoding section (xvid/x264 or whatever you are using).

[Gladiator:)]

Ya i am using XviD

Pl someone suggest this

[unskinnyboy]

The encoding speeds seems normal considering the filters you are using. However, you should post the whole of your current codec settings, and your script for us to suggest speed improvements (beware - most speed improvements will reduce quality!). If you have a dual core, you might try multi-threading Xvid and use MT Avisynth.

[TheRyuu]

I doubt the problem is xvid itself. Make sure you have the latest version. Celtic_Druids website has some later builds.

Post your script, could help out in determining whether or not the problem is really from xvid or the script.

I have a Xp 2.66 GHz and a 759 MB ram speed

This tells me nothing of what kind of processor you really have.

[Comatose]

Athlon XP 2.66 GHz, eg a single core.

If you use a really slow avs script, it might be wise to encode to Huffyuv (single core CPU) or Lagarith (multi core CPU), which are lossless codecs, and then encode to xvid or whatever from that.

[Blue_MiSfit]

Yeah, your CPU is most definitely the bottleneck. If you're unhappy with your speed it's time to upgrade. Core 2 Quad is the way to go :)

Encoding to lossless first definitely helps for the slow scripts, but gobbles up space!

Q6600 is a great deal.

~MiSfit

[Gladiator:)]

Yeah, your CPU is most definitely the bottleneck. If you're unhappy with your speed it's time to upgrade. Core 2 Quad is the way to go :)

Encoding to lossless first definitely helps for the slow scripts, but gobbles up space!

Q6600 is a great deal.

~MiSfit

First i have to encode into lossless with 2 or 3 pass ?
Then while again encoding into favourable size i have to again use the same script which i used for lossless or leave it blank ?:confused:

[Blue_MiSfit]

Encode your script to (for example) Lagarith YV12 using VirtualDub or whatever you want. Lagarith is lossless, and will store the precise output of your script. It will be large.

Then, instead of loading the script for your first and second Xvid passes, load this Lagarith AVI (or a simple AviSynth script with AVISource(...)).

It can be hugely faster to simply read and decode a lossless file than it is to load and execute a complex filter chain. Especially when doing two passes.

~MiSfit

[MadRat]

Blue_MiSfit is right. If the filters are slowing you down first encode to a lossless codec. There are two main choices of codec. If your hard-drive is short on gigs use Lagarith, it's slower but not too badly so and will give you a slightly smaller file size. If you want speed use ffdshow's HuffYUV, though the file size will be larger. On my AMD 64 Anthlon 3200 and Windows XP 32bit I get between 35 and 40 frames per second when I encode a ffdshow HuffYUV AVI file.

Here's the logic behind the strategy. Let's say it takes uh... I dunno... 10 hours for each pass because of the filters. So when you're encoding two pass with xvid it takes 10 hours per pass plus the encoding time of another say... hour and a half for 21 and a half hours total. If you encode to a lossless codec it will take 10 hours plus the hour and a half xvid two pass encode for a total of 11 and a half hours. That saves you 10 hours of encoding time.

Why do you a lossless codec? Each time you loose a little detail when you use a lossy compression method. You can try this experiment. Take a picture and save it as jpeg, then save that picture as jpeg, then save that file as jpeg. After a while you'll end up with the picture looking like a bunch of big colored tiles. It's the same idea as photocopying a photocopy or video taping a video tape or faxing a fax over and over. Lagarith and HuffYUV are more like compressing with Zip or RAR. No matter how many times you compress and uncompress the file it comes out the same.

[Gladiator:)]

Ok now what should i do
Pl don't take it too technical i am new
First can i convert into MP4 lossless using MeGUi Itself ?
How disk space it needs ?
Then input the Lossless and encode into AVI rite ?

[thetoof]

No.

Load your script in VirtualDub (or vdubmod) and do a "Direct Stream Copy". This'll output a Lossless YV12 .avi file on which all your filtering will be applied, but no compression. Then, create a script with avisource("mylosslessfile.avi") and load it in megui to do whatever you want. This way, your filtering will only be computed once and it'll speed up the overall process.

The file size should be around 450Kb x number of frames. I have a 2h18 movie that takes 86,1 GB on my hard drive.

edit: that was for a "normal" YV12 output. To use lagarith, select "fast recompress" and then lagarith in the "compression" menu (ctrl-p). You may have to install it before. (http://lags.leetcode.net/Lagarith_1315.zip) The file size should be around 4 times smaller than what I mentioned before. In the config window, make sure to select "mode=YV12"

[Blue_MiSfit]

Whew! I was scared you were going to make the poor guy save uncompressed YV12!!!

~MiSfit

[Gladiator:)]

Is that any way to encode it with >18 GB ?

[Ranguvar]

> 18 means more than 18... I think you mean less?

And probably not, if you want to do a lossless encode first. You'll either have to do a high bitrate lossy first then, or just sacrifice some speed and do it right from the script.

[Gladiator:)]

ya less space