Best settings for "The Matrix"?

[w00d]

I'm getting ready to do a 1-CD encode of The Matrix (to watch on my laptop, no DVD drive) and was wondering if anyone could recommend some settings to use? The reason I ask is my computer is slow (466mhz Celeron) and it takes me all night just for 1 pass, so it's really hard to experiment and do lots of configs to see the best result.

I know that The Matrix is very compressable since it's dark and lots of the same colors, etc. so I'd probably get good results using doom9's settings in his XviD guide but I want to get the best looking rip. So if anyone could give me some pointers I'd be very thankful.

BTW I've already encoded the audio using vorbis and that leaves me 604,000KB for video. If that helps you at all...

Thanks :)

[Acaila]

I've done a 1-CD rip of The Matrix a while back using XviD and Ogg Vorbis as well. I tweaked it for all it was worth, but it never came out to my satisfaction (or maybe my standards are just too high :D). The movie has just too many sudden high motion scenes and the codec won't respond quickly enough, so you'll get macroblocks all over the place.

[Teegedeck]

I can't imagine that the codec doesn't 'respond' quickly enough - anyway, I'd use low aggression on this one with wide ranges, that should give a more or less constant quality, shouldn't it. About lumi yes or no, I'm not so sure. I'd tend to say it would be useless in a movie that is that darkish but I'm really not sure.

[sierrafoxtrot]

shot in the dark.

lo% 0
hi% 250
high aggression
50% strength

i'm thinking that high aggression will allow those large frames to be compressed less, because low aggression tends to steal from the large to give the the smaller frames. hence, setting lo% at 0 will scale the curve so that small frames aren't hit too hard. good luck ;)

[rui]

Originally posted by Teegedeck
I can't imagine that the codec doesn't 'respond' quickly enough

I had experienced this also, when i tried to encode a rock concert from a rock band called "Pantera".
It's hard rock, with lots and lots of movement, both from the artists and the audience.

When the artist starts jumping around the stage, there are lots of macroblocks for a fraction of a second, and almost immediately the codec "adjusts" and the macroblcoks disappear.
Then, when the artist starts jumping again, again i get lots of macroblocks, and almost immediately after that the codec again adjusts the encoding and the macroblocks disappear.
But there is always a moment in time preceding the most agitated scenes where the picture gets very nasty.

[The People's Elbow]

@rui: Do these macroblocks maybe appear only in keyframes? If that's so, than try to increase the kf boost percentage. (Though 20% is a well working value imo)

greetz, Elbow!

[w00d]

sierrafoxtrot, I encoded the movie today using your settings and it came out great. The overall quality is very good and there are hardly any macroblocks to be seen. Thanks! :cool:

[scorchED]

wOOd:which resolution did you choose?

[w00d]

640x272

[rui]

The People's Elbow, thanks for the tip. :)

I am using the alt. CC, and i leave both the keyframes and p-frames in the 2-31 range. Maybe if i restrict the key-frames in 2-4/5 range, it could better things?
By the way, do you know of any way that i can tell, when playing the resulting avi, if the problem arises in key-frames? Is there any software (debugview is only for the encoding process, right?) that would show me that?

[[Toff]]

If this is not a DirestShow problem you can look at key frame with VirtualDub.
Open your movie in VirtualDub,
use Shift + Slide bar to navigate on key frame only.
Also you see a "K" if the current frame is a key frame above the status bar.(ex: Frame 600 (0:00:24.000) [K])

[rui]

Thanks for the tip :)

[The People's Elbow]

@rui: [Toff]s' method is what i was thinking of... load your file into vdub and check the I-frames and compare them to the following P-frames. If the macroblocks are stronger in the first ones then the "I-frame boost %" should be increased. This setting is in the two-pass tab @Two-pass tuning... I won't recommend to lower the quantization range for the i-frames though I never really gave it a shot. The boost option works great for me.

greetz, Elbow!

[rui]

Originally posted by The People's Elbow
@rui: [Toff]s' method is what i was thinking of... load your file into vdub and check the I-frames and compare them to the following P-frames. If the macroblocks are stronger in the first ones then the "I-frame boost %" should be increased. This setting is in the two-pass tab @Two-pass tuning... I won't recommend to lower the quantization range for the i-frames though I never really gave it a shot. The boost option works great for me.

greetz, Elbow!

Elbow, thanks to you too :D

But for i to have access to that setting (i-frame boost), i have to use the normal curve for 2nd pass, and i used the alt. curve compression. So, the only way i have to up the i-frame quantizers is by limiting their range, correct?

[The People's Elbow]

is the i-frame boost disabled if alt cc is enabled? There is no real relation bewteen I-frame boost and moten compensation, but the same tab in the gui, so... it would be logical if this function is not depending of the cc type and enabled all time.

greetz, Elbow!

[Teegedeck]

I-frame-boost is enabled when using alt.cc also. Just watch dbgview; why do you think I-frames receive noticeably lower quantizers? :) (Heh, I'm glad we got that feature, I was always bugging Koepi and -h that I didn't like the 'Nandubization' of XviD, with quantizer-ranges and all that.)

[Shayne]

The matrix 40 meg (12968 frames) file scaled for full movie 610 meg

Avi's compiled

Straight

M_200_100_50.avi No notible improvement

M_200_50_50.avi Better

M_200_25_50.avi not as good as above

M_200_75_50.avi not as good as 50

M_200_60_50.avi bit Better than 50

M_200_55_50.avi almost idential end of low distance

M_150_60_50.avi not as good as M_200_60_50.avi

M_250_60_50.avi almost identical end of high distance

end result low distance 60 high distance 200

H_200_60_50.avi a bit better we are talkn real small now

H_200_60_50 mod.avi no great increase take the hint

H_200_60_50 luma.avi not much change if any

Conclusions

Encoded full movie no atl code and H_200_60_50 end result: stills and slow mo a bit sharper crisper with the alt
If you have to tweak this for every movie the time, unless u have it, is not really worth the results

My feeling is the codec should work for u not the other way around. We have 610 meg and 140 minutes of video compression ratio/quality is codec dependant and i wonder if curve compression is the miricale dvd quality fix?

[rui]

Well, in answer to your and, most probably others, desires, Koepi just spoke in this thread about in the future xvid will have the alt cc implemented in automatically, without user input :)
Look for the last posts.

http://forum.doom9.org/showthread.php?s=&threadid=19071

[c0p0n]

Originally posted by Teegedeck
I can't imagine that the codec doesn't 'respond' quickly enough - anyway.

I experienced such things on Mike Oldfield´s Art In Heaven DVD concert, specially on the final part (Berlin 2000) wich shows the light show - I had to tweak that a lot to avoid huge macrbolocks to appear.
high aggression, hi 250, lo 84, strenght 50