It's been awhile since there's been anything said on compression here. Looking for answers I thought I'd find them all hidden here and through Google's old cache of the Virtual War Cinema... While I've found a lot of good information, I'm still left with questions. Some threads are so old I don't even think some of the settings people were using are even options anymore! Please do not direct me to guides and tutorials. Settings that work great w/ real life video do not necessarily work well with IL-2 video. While there are many threads at Machinima.com, I can't help but take them with a grain of salt because the sheer image quality of the movies there is no contest to some of the movies that come out here. Also, the standards here are much higher and people are willing to download a larger version of a movie just to get the higher picture quality.

Yes, I can spend hours and probably whip out something halfway decent by trial and error, but I figure I'd ask the source and possibly find out what is already working and possibly crank out something a lil better! I often times find responses from people referring to things like the # of passes when talking about a certain codec, yet I don't quite understand exactly what they mean. Do they mean multipass, 2-pass, Nth pass? There is no 2 pass in divx and there's 2 different types of 2 passes in xvid? The responses hardly ever go into enough detail.

So I ask the movie makers, exactly what settings are y'all using and what advantage/disadvantages do you think they give you!?!


DiVx 6.4 (current version)
Rate control and bitrate:
1-pass, quality-based
1-pass
Multipass, 1st-pass
Multipass - Nth pass

Codec Performance
Frame Control
Motion Estimation
Image Processing



XviD Mpeg-4 1.1.2 (Current version)
Single pass
2 pass - 1st pass
2 pass - 2nd pass

Profile @ Level
Target Quantizer
Zones

Hi Direwolf,

The compression for POW2 was via VirtualDub with the following settings:

Audio Compression:
Lame MP3 48,000Hz ? 128kbps CBR Stereo

Video Compression:
Xvid MPEG-4 Codec
Profile Level AS@L5
Encoder Type: Single Pass
Bit rate: 1250 kbps (Compressed file size was 268mb)

All other compression settings kept at default

Hope this helps

I'll tell you what I do, others can chip in.

This topic could really do with a decent description by someone, then stikied, as it's asked so many times.

I'm still on Divx 6 (I ain't changing until FH&C is finished)

Load movie in VirtualDub, click Video, then Compression
Select Divx
Click Configure
Click restore defaults - to clear up any rubbish
Then Home theatre profile
Then Multipass - 1st pass
Then 1250kbps
Save the movie
DO NOT EXIT VIRTUAL DUB - click Video, then Compression
Select Divx
Click Configure
Then Multipass - Nth pass
Save the Movie - it will say 'do you want to overwrite?' - say YES
That's it.

Don't forget to compress the audio - I use mpeg3, 128kbps, stereo

Fantastic Monty & Joe!

Exactly the type of responses I was looking for! No vagueness, no bs. Thank you so much!

It's funny you said that Joe because my ulterior motive was to get something stickied for the nOObs
Nothing is more frustrating than searching forum to forum trying to figure out who used what on what.

I'm interested in the divx settings, I've been playing around with it the most lately and everytime I use Nth pass I get only audio and a black picture.
I've never had much luck getting good results with xvid, but then again, I have never really known where to start so I can't wait to try that too.

Thanks guys!

Joe, may I ask, why do you compress twice at 1250kbps? This would heighten pixelation and blockiness in my opinion. You are taking a 1250kbps file and re-compressing it again at 1250kbps... any pixelation and banding introduced in the first compression cycle would be enhanced through the second cycle, that seems fairly logical no?


MrVH

I know what he's going to say.... and ironic because it was one of Joe's posts from old I found that said he did 2 pass divx. (Which didn't make sense to me because I couldn't find a setting that said 2 passes!)

Multipass-1st pass is just that, the first pass. It does not do a 2nd pass in that op. The second pass is actually Multipass-Nth pass, (Which explains why I always get audio w/ no picture when i do that, I never did a 1st pass). Divx you can run as many passes as you'd like. Xvid just has a setting for 2 passes, no more. If he's running a second, it's to squeeze just a little more off the size w/ minimal loss of quality.

My source:
http://www.divx-digest.com/articles/divx6_setup.html

How'd I do?

MrVH, the answer is related to Direwolf's Black picture.

The first pass only analyses the video to determine the best way to apply the codec algorithm. It only stores a logfile, and no video - hence a black screen.

Then Nth pass takes the logfile from the first pass and uses that to improve the algorithm of the actual video compression.

I was helped in a very big way by Joe90 and 310th Falcon and i will ask Falcons permission here if i may add his response to me which can be included in the sticky ?

Direwolf19 i was struggling to compress movies and was getting **** quality and high Mb before i was pointed in the right direction and now after 6 months (yes 6) my movie on the RAID OF PLOESTI is due out shortly (little plug).
Thank you guys you all never tire of helping us nOObs

I should probably reply to my own thread sometime huh?

Deadliest Enemy was done in 1250kbps Divx 6.2.5 w/ 3 passes (1st and then 2 Nth's). I had versions w/ 4 and 5 performed but I really didn't see a difference and the size was the same. I had no luck w/ Xvid, I tried MontyDan's settings as well as a few I found from other boards but it just wouldn't work right.
Audio was Lame Mp3 48,000Hz, 128 CBR Stereo.

All of this was done in VirtualDub from a giant uncompressed clip. Adobe Premiere 1.5 was used initially until I upgraded to Premiere 2 about 80% through the project.

you should always remember, that the most important factor is bandwidth. so, if you already use very high bandwidth, the effect of 2-pass compression is less visible. however, 2-passes (or more) are of big advantage, because it is the only way to achieve variable bitrates - that way scenes with motion get more than still pictures.
direwolf is very right by saying, that different footage requires different settings. in case of game-captures the most influence hereby has the absence of motion blur. i regard bitrates from 850kbps (~640x480) on upwards as very high. starting at 1500kbps on i would prefer mpeg2 at normal resolutions.

here are some basic settings for 'xvid' in 2-pass compression:

when using 'xvid' for the first time, better press 'load default' once.
the first tab shows like this:



chosing 'Advanced Simple @ L5' press 'more' (1) to see this tab:



'Quantization type' should be 'H.163' for desired bitrates below ~1000kbps and 'MPEG' for above.
'Adaptive Quantization' is meant for saving bitrate in scenes which are hard to see anyway (like very dar ones).
'Quarter Pixel' are more accurate motion estimation.
'Global Motion Estimation' is more accurate on high motion/panning scenes.

after clicking 'ok' and back on the first tab, chose at 'Encoding Type' the 'Twopass - 1st Pass' and with 'more' (2) you can designate the location for the logfile.
at 'quality presets' and 'more' (3) you can tune it some more:



the 'vhq mode' makes only sense with 'GMC' chosen at 'profile @ Level' before, else keep it at '1'.
the 'Maximum I-Frame Interval' should be around 10times the framerate.

on the second tab of 'qulity presets' keep all as it is (if you don't want spend very long time for testing), checking 'Trellis quantization' is ok though:



after done all this, you can start the first pass.
when it has finished, go to the compression settings again, and just change 'Encoding Type' to 'Twopass - 2nd Pass' (maybe check under 'more', if the location of the logfile is correct) and chose the desired bitrate (or the desired filesize) by moving the slider or typing a value in:



start the second pass, and - if you go for even better bitrate distribution) - a 3rd or more passes (the effect will be less and less visible though)


this is just some general setting for the start, but there are a lot more possible changes of the quant-matrix possible than in 'divx' (for enthusiasts )
i prefer 'xvid' by far over 'divx'... i find the quality at same bitrates visibly better and the encoding is a lot faster. another reason is, that 'divx' (since the introduction of it's own containerformat with v6.x) has shown to give some major problems and incompatibilities with other codecs and filters. in special there is any kind of subtitles using 'vobsub' for filtering (which are not to be seen or have green framed pictures), furthermore random colour-aberrations with other mpeg4-coded videos ('qt-mp4' and 'xvid'), and - pretty nasty - the codec writes its merits back to front with every time of use (so it insists to decode all possible mpeg4, despite changes by user). the last issue seems to be fixed with latest version of 'divx'.


edit:
to answer direwolf's question for 'zone' and 'target quantizer(?)' (do you mean 'target bitrate/filesize'?)...
the 'zone'-option lets you define parts in the video, which you want to give less bitrate (quality) than the others (count in frame-numbers, quality-factor as weight). this may be useful for titles, fe. in general you will use it for compression of cinematic releases, where you can save bandwidth at super-long endtitles.
the 'target bitrate/filesize' toggles between these two and adjusting the slider will let the inbuilt calculator estimate the respective other value. this is useful when, fe, compressing to the size of a cd (700mb) -> the calculated average bitrate is in general very exact (+/- 1mb)

This message has been edited. Last edited by: deepo_HP,

Thanks deepo, I've always been to scared to play with all those settings, but I'll use your advice next time.

(Tooz, this is developing into a good sticky?)

BUMP - can you please sticky this thread?

Hmm maybe this stuff can help me on my next video. Even though I'm an ultra noob and don't know anything about compression, etc. All I've done is record with Fraps and save with WMM.

I hope to make a movie for my grandfather someday. Or maybe even if someone is willing...help me make one. I think I have a good story for a video.

Great thread guys. If I had anything to add I would but alas 'm gonna stay well out of it.. well apart from there was a great scene in POW that required special attention due to pixelation and that was due to the similarity between sky, plane and sea colour!!!!

It would be nice to see some real life figures on what it can physically do to a movie. ie. nominal single pass file size at 1250k and then subsequent file size variences when using nth passes!

Bumpity,

Old but still relevant - just like some of us in here

quote:

Old but still relevant - just like some of us in here

here...here

This should be put in the "sticky" department...


Best Regards

hi all,

since we are at it with bumping

on re-reading this thread, i thought it could be good to get also bbloke's question discussed.

quote:

Originally posted by BBloke:
It would be nice to see some real life figures on what it can physically do to a movie. ie. nominal single pass file size at 1250k and then subsequent file size variences when using nth passes!

i am not really sure, what you mean with 'real life figures', bbloke?
however, the file-size is always and only dependíng on the bitrate, not on passes used. compressing with 1250k should always result in, well, in 1250kbit/sec - or 9mb/min.
compressing in 1-pass this will be achieved at 99%, using more passes the bitrate is averaged, so it can result in slightly different size, though should be still at 98-99% accuracy.
the advantage of multi-pass is, that the quality at the same bitrate (=filesize) will be better, as the available bitrate is not constant, but variable, giving scenes more bits/pixel, which demand more (like lots of action) and less bits/pixel to scenes which need less (like slow motion or still pictures).

this might sound the same, but the multi-pass doesn't decrease file-size, it only allows to encode with lower bitrates and still getting similar quality. however, some testing with small scenes is necessary to see how much lower one can get.

(it might also be what bbloke was thinking of ... i guess so. i found it worth pointing out though, that the file-size is always determined by setting the bitrate - multipass is just the 'robin hood' of digital movie: taking bits from the sitting fats and giving it to the working actions)