View Full Version : DivX5.1 3-pass compared to 2-pass
DarkWave
11th September 2003, 11:32
Hi,
there´s just a little question from me. I compared a 2-pass encoding with DivX5.1 to a 3-pass encoding.
And what should i say? Quality is nearly the same except of one huge problem: there are lots more macroblocks in the 3-pass than in the 2-pass.
My settings
2-pass:
GKnot, lanczos, Resolution: 640x352 Pal, Filmlenght: 0:42:25, File Size: 350MB (to get 2 files on one CD), Bitrate: 925KBit/s, 1st/2nd standard/slowest, no mv, pv low, b-frames, no gmc, no qpel
3-pass:
VDubMod, same settings as above, 1st/2nd/3rd standard/standard/slowest, no mv, pv low, b-frames, no gmc, no qpel
Anything wrong with that settings?
DarkWave
temporance
11th September 2003, 12:10
Did you accidentally use the output of the 2nd pass as input for the 3rd pass?
DarkWave
11th September 2003, 12:32
what i did, was the following:
- i converted the dvd2avi output with VFAPIConv so vdub could handle it
- edited the params for 1st-pass (btw with 4000KBits/s), added it as film.avi to the job-queue
- edited the params for the 2nd pass (920KBits/s), then again added it as film.avi to the job-queue
- 3rd pass, same as 2nd pass - changed only standard to slowest mode, and again, saved it as film.avi
- then i opened the jobs-menu and started the whole queue...
was it wrong to use 920KBits/s in 2nd AND 3rd pass? must i only use it in 3rd pass and give 2rd 4000KBits/s?
DarkWave
kl33per
12th September 2003, 07:28
All passes should use the same bitrate.
DarkWave
12th September 2003, 08:21
Originally posted by kl33per
All passes should use the same bitrate.
sure? in all guides you find informations, that the 1st-pass uses a higher bitrate than the other passes...or did i misunderstood these guides?
N_F
12th September 2003, 16:45
Originally posted by DarkWave
sure? in all guides you find informations, that the 1st-pass uses a higher bitrate than the other passes...or did i misunderstood these guides?
While the codec does indeed do something like that, it'll do it automatically without your input. You should use the same bitrate for every pass.
DarkWave
12th September 2003, 17:15
Hi,
@kl33per & N_F
thx for your comments. for me it was clear to use the same bitrate in the nth passes, but ´cause of the possibility to enter a value for the first pass, i thought i had to do it manually.
but the old problem is still there: macroblocks in a 3-pass encoding, even when using a denoiser (source is a bit noisy - stargate sg-1 1st season box).
3-pass leaves less artifacts around people or objects, but produces more blocks, especially in scenes with a high amount of blue colors...
EDIT: you can find some screenshots here... (http://members.lycos.co.uk/muliekino/sg/)
DarkWave
Acaila
12th September 2003, 18:00
Actually AFAIK when doing the first pass, the codec encodes it with the bitrate you specify, not with 'extra high quality'. That way it will gather the complexity/motion information for the correct bitrate so the 2nd pass won't deviate too much from the known info.
GordianKnot on the other hand performs a compressibility test with a constant quant 2 file, not with a normal 1st pass file. Also, XviD's 1st pass is done with a constant quant 2 encode as well, not bitrate mode like DivX.
N_F
12th September 2003, 19:49
I've been away too long... I had XviD in my mind while writing that.
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.