Announcement

**Dr Mordrid** · 1 July 2000, 14:08

Fast and slow are both variable bitrate but there is a big difference in how they respond to the bitrate settings. Don't think of their names in terms of the motion compensation but instead in terms of processing speed.

Fast will rarely go over 500-1000 kbits even if set for 6000 kbits. It tries to keep the filesize as small as possible.

Slow will stick pretty close to your settings, varying up and down according to the action. It's target is quality.

You might also want to check this page. There's a lot of good stuff in the Articles page;

http://www.digital-digest.com/divx/index.html

Dr. Mordrid

[This message has been edited by Dr Mordrid (edited 01 July 2000).]

**Enrico Ng** · 1 July 2000, 14:28

yeah I went to that page before.
it said that I should use low motion and set crisp to 0.

when I try fast with crisp at 100 and bitrate at 2000 it is still choppy, and 28MB by MPG is 33MB, the whole point to me converting to DiVX was to save HD space.

I would expect that I would be able to convert a MPG1 to MPG4 with out losing quality but with a file size about 25-50% of the orig. not bigger

**Enrico Ng** · 1 July 2000, 18:59

I use the fast codec to get the smaller file sizes, usually using the fast codec with a bitrate of 2000 will work fine, but the file size will be about the same as a MPEG1.
I can get the same results with the low motion codec, but the file size will be even bigger.
how do the people that rip out movies like the matrix manage to fir a 2 hour movie in 600MB and still keep it smooth and clear.
The best I can do is 400MB for 40min of video

**intomisery** · 1 July 2000, 21:58

Enrico, try one of those bit-rate calculators that are on the page listed above. They work quite well. Also, if you're using FLASK, make sure the deinterlacer is on and make sure that you check the non-MMX IDCT decoder. In my experience, you get far better results with that one. if you can affor the time, choose the IEEE IDCT decoder.
Yet again, scale down the video. I have learned that using a smaller resolution allows the datarate to provide better quality as it has to work with a smaller area. 1000 kb/s over 640x400 will give you better picture than 1000 kb/s over 720x480.

Hope this helps! have my own fun (sarcasm) with DivX at the moment.

**Enrico Ng** · 2 July 2000, 06:06

bitrate calculators are pretty useless for two reasons.
1. they cant get an exact number
2. it doesnt matter, I want great quality above all else, the only problem is that the only way I can do that is the make the filesize big. almost the same size as an MPEG1. (makes sense). I just dont understand how those "rippers" can make matrix with great quality and make it so small.
ie. matrix is an average of 4.5MB/min
my videos are an average of 15mb/min

**Tinnes** · 2 July 2000, 08:26

Umm I don't know much about Divx either and I have just started playing round with it.

Anyway I was just wondering if you are compressing the audio aswell? If not then use VirtualDub and load up the .AVI file that Flask chucked out. Set the video to 'direct stream copy' and audio to 'Full processing' or something like that. The select audio compression and use MP3 on 'MPEG Layer-3,95 kBit/s, 44,100 Hz, Stereo'. Then just save it to a separate .AVI file. This GREATLY decreases the file size.

Sorry if you are already doing that.

Tinnes

------------------

Get paid to surf http://www.alladvantage.com/home.asp?refid=MAC649 it really works. Free Money.

**Joxx** · 2 July 2000, 08:41

Enrico,

I've compressed the Matrix with DivX down to 640MB. Let me give you some advice.

First, don't use the FastMotion Codec. The Fastmotion codec doesn't really use 2000kbps bitrate even if you select that. The FastMotion always try to go down to 500kbps which will make the file smaller but more unpredictible. The bitrate calculator only works with the LowMotion codec. Use the lowmotion codec.

Then we go to sound. You could save bits for video by lower audio bitrate. But I cannot recommend that. Audio is very important. You can look at bad video without notice, but you can't listen to bad audio without notice. That's why Real has a higher priority on sound than on video, It is better if video than sound is lagging. In the VCD format audio has very high quality while the video isn't taht good. By decreasing audio bitrate to half, you only gain a few kbps on video bitrate. Especially in a action movie like the Matrix, audio should not be compromised. Use atleast 96kbps, or recommended 128kbps audio (mp3 or WMA).

Then video. We have the Matrix, 131 min long. With 128kbps audio we have 550kbps of video bitrate. To increase the bitrate you could compromise audio (not recommended) or cut the last five minutes of text. Actually the text has a function and the Matrix won't be the same without it. We'll come to that later. You could also use 2 CD-R which will doubble the bitrate. But with a bitrate of max. 600kbps the DVD size of 704*576 (PAL) is absoloutly impossible. To acieve at least low quality you need 1200kbps, probably 1600kbps. I resized it down to 352*288 (PAL). This format is very small and will give the movie (on television) a blurred look, but it is at least VCD size.

Increasing video quality even more is quite complicated. Workstation-hardware encoders used for encoding DVD video uses two-pass encoding. This is a very good technique to optimize video quality. The codec goes through the video once (pass one) and check where the bits are needed to give optimal quality/size ratio. The second pass is the actual compression wich will encode the movie with the bitrates given from the first pass. DivX uses variable bitrates which also can add bits where needed but not with the same accuracy and performance. The bits will only vary a little so that the given bitrate will be accurate. What you can do is to do the first pass yourself. Check where bits are needed.

The ending text in the Matrix doesn't need a very high bitrate. Cut those 5 min. and encode it with 150kbps (I've got this bitrate by testing), you will acieve the same quality since a higher bitrate isn't needed. Then cut together the lowbitrate clip with the rest of the movie. This will save you some space and higher quality/size ratio. You can do the same thing with other lowmotion scenes. And on the other hand you can encode highmotion scenes (which will have low quality and show attentionable artifacts with 550kbps bitrate) with a higher bitrate. This will give you significantly increased quality.

But the best and most easy technique is the two-pass encoding. I hope that we will have a software 2-pass encoder soon.

I hope it helps.

**Enrico Ng** · 2 July 2000, 09:48

I understand the diff between low and high codec. as I understand it, high will keep the bitrate low for low motion and raise it for high motion scenes, that way, it saves file size. ofcourse this isnt as good as low motion, but with lowmotion, it keeps the bitrate realativly high in low motion and fast motion. that is why I usually use high.
whenever I use low, I alway end up with a bigger file size.
the videos I am encoding are about 15min long at the most. these are video I capture with the marvel. the specific vcd mpegs I am trying to convert to divx are from another source.
in either case, low or high motion codec, I still end up with a large filesize. converting to divx from mpeg1 seems useless because the file size is about the same.
I usually use 2000 high.
when I ise low motion I set it around 700-900 to get the same results.

[This message has been edited by Enrico Ng (edited 02 July 2000).]

**cjyo~** · 2 July 2000, 10:27

Joxx,

What do you use to cut and splice your different bit rate clips together (Virtual Dub)?

**Enrico Ng** · 4 July 2000, 18:24

I hae the matrix its 2hours 16 min and around 600MB, its not choppy at all. how did they do this? for me, the lowmotion codec doesnt work unless I bring it up to the same datarate as for the fast motion codec, otherwise its choppy.

I dont get why the lowmotion codec doesnt work for me and how those people made matrix

**Joxx** · 5 July 2000, 14:28

cjyo,

virtual dub, open clip one, then open second clip with "append segmented file" or something, it is the second alternative in the file menu. Then you have the second clip at the end of clip one. If you want to ha. ve the second clip in the middle of clip one, cut the end of clip one by selecting the frames: skip to the start and end frames of the series of frames you want to cut and select them with "home" and "end" the press del. If you want another clip at the end simply append another segment and cut the unwanted frames.

to append a clip, the clips must have the same size and framerate. Possibly another, for me unknown, criteria too.

**Joxx** · 5 July 2000, 14:41

Enrico Ng,

you have to understand that the size of the avi is not randomized. The size is simply the sum of audio bitrate * length and video bitrate * length.

A 60min (3600s) DivX file with audio bitrate: 96kbps and video bitrate: 910kbps cannot be anything other than 3600(96+910)=3621600kb (452,7MB) big.

I don't know how you do to get that big files but it is not a very advanced thing. If you check the properties of you file, and calc. length * (audio + video bitrate) you will find that the rules are accurate. If not, post back on the forum.

Perhaps you havn't encoded the audio?

**Enrico Ng** · 5 July 2000, 17:05

i thought you said that divx was a variable rate codec

**Joxx** · 6 July 2000, 06:49

Yes it is, but the variation is limited. It does not vary more then just a few percent. If you select 200kbps you wont get 3000kbps, but maybe 230kbps.

So, you are correct, it is a variable encoder. But a very restricted one. Otherwise, the encoding would get out of control and you would always get very large file. The positive and negative variation is intended to be equal. Of course this is not always the case, but if it is no negative variation it will not do positive variation either.

Announcement

MPEG->DiVX

MPEG->DiVX

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment