View Full Version : Soft bicubic@60% first pass or neutral bicubic@50%?
soujir0u
11th June 2002, 13:18
Hmm, I ran two first passes for my movie with Soft Bicubic and Neutral Bicubic (using Avisynth) and I got 60.3% and 51% respectively. This movie is a real pain to encode, I was using 512x288@29.97fps. Problem is that all the frames are progressive, and if I use Force FILM or Decimate (Decomb), the video gets really choppy so I have to use all the frames. Using temporal smoother didn't help either (1% advantage only).
Anyway, which resizing would you guys recommend in this case?
I would choose the 512xXXX and activate the lumi masking, ahh and the credits in Xvid. may be make a small audio etc...
Also you could change manually the .avs and modify the b,c parameters of BicubicResize to something like 0.1,0.45 and test, but don't forget that this blur the image a bit, the lower b,c the blurrines.
Try and post...
/Bypass the grammmar and etc.. etc..
soujir0u
12th June 2002, 06:34
I decided to just stick with Soft Bicubic, Neutral gave a lot of high quants (bigger than 7)... Lumi Masking helped a bit, but I noticed some minor blocks so I turned them off cause the compressibility advantage was very small (about 3%).
Malinka
12th June 2002, 09:43
Sorry in advance for abusing this thread :rolleyes:
I am trying to get Xvid work for me. So I was wondering how to make a compressability test with Xvid.
For my last encoding attempt I made a first pass and then loaded the stats file into gordian not using the load button in the bits/pixel*frame section. Is that right? And, which percentages do give decent results?
Thank you very much in advance
I'm sure if you use the find button you could find info about it,
but here we go anyway...
'Couse XviD is more bitrate-efficient you could point to 65% or lower may be 55%.
GK is a good tool for that, also you can use quick2pass that can make a 5% 1pass and give you a com. test
and nice to see other Girl in the Forum!;)
Malinka
12th June 2002, 22:16
Sorry, you are right. Affirming the clichés about women one more time :( I will try the search button next time. Big Promise! Anyway, thank you very much for your answer. :) :)
It is strange however. For my last encode was encoded with 70% or more and still showed a fair amount of artifacts and loss of detail. Especially on finely structured tapestry or stone walls, the macroblocks somtimes lost every detail turning flat and monocolored. :confused:
:) :) :) :)
soujir0u
13th June 2002, 02:20
Hmm, 70% should give pretty good quality. Did you turn on any post-processing options during playback? More smoothing = less details.
Btw, Kyo = from KOF?
@soujir0u
Yep Kyo from KOF, Ore no cachi da! (in English some like: The Victory is mine)
and Tenkken Soujir0u of the Ten Jupponggatana in Rurouni Kenshin ? :)
@Malinka
You have played with the values on alt cc. or normal cc.?
BTW. what movie, Resolution, audio, DVD, Country, Tel., Address .... :D
We need more info :p
/Escuse me the typos and the English, oh also the Japanesse :eek: /
soujir0u
13th June 2002, 04:34
Yup! I love Soujiro!
I like Kyo too, but I like Iori more!
Ioriiiiiii....... :)
Kenshin is one of my first-and favorite mangas/anime!
Malinka
13th June 2002, 10:30
The film was Woody Allen's Mighty Aphrodite (I guess I don't wanna hear any comments about my taste ;))
97min, PAL
I was going for a video-file size of 646000kb,
Quantization Type: Modulated (Pass1&2)
Max I-Fr Interval: 600 (movie has long scenes)
Lumi: yes (Pass1&2)
MinQ I-Fr: 2 I-Fr Boost: 5
MaxQ I-Fr: 6 CC High: 10%
MinQ P-Fr: 2 CC Low: 10%
MaxQ P-Fr: 8 CC Payback: 500
Hinted ME: yes CC biased Payback
Alt CC: No
Avisynth:
# CROPPING
crop(8,16,704,544)
#
# DENOISING
TemporalSoften2(2,3,3)
#
# RESIZING - sharpBicubic
BicubicResize(544,288,0,0.75)
soujir0u
13th June 2002, 11:11
Hmm, perhaps you've filtered it too much with the temporal soften. Also, I remember that using Modulated on first pass gives MPEG quantizers, so you might want to just use MPEG for both first and second pass. Neutral bicubic should be very sharp already, you might want to try that instead of sharp bicubic to increase the video compressibility.
serbersan
13th June 2002, 22:46
A little advice Malinka.
Decrease resolution of your video instead of use a filter like temporal smoother or temp soften.
You will gain in speed at least. It isn't really recommended to use a filters unless you know perfectly what are you doing. And not only me, very very good rippers like doom9 doesn't use filtering (like temporal smoother or any smoother filter...), never.
And if you want to use modulated use first h263 in first pass and modulated in second.
If you really want to use Filters, to get the maximun quality without blur or remove details in the image, i recomend TS(1) or in extreme situation TS(2,1)
If you use lumi in the 1&2 the efect is DOUBLE, i only use it in 1st pass.
using hinted me speed up the process but the quality might be touched a bit.
Also i would choose a higher i-frame boost between 10-20%
-Kyo
vBulletin® v3.8.4, Copyright ©2000-2009, Jelsoft Enterprises Ltd.