CCE bug with avisynth ONLY

[gurgenpb]

When using the avs script and link2 to make a fake avi, CCE behaves very strangeley. My setting is 3 pass VBR (which is as I understand the vaf and 2 separate passes). Well, I get the vaf file OK, then the first pass .m2v file OK, followed by the second pass .m2v file that is exactly 4/5 the size of the first pass .m2v file (which is of the correct size!). So, why am I getting a smaller file size on the second pass only. I am encoding film-based dvd content.

When I use VFAPI instead of the avisynth script, the file size comes out perfect!

The most curious thing: the 4/5 file size is of the exact ration of the 23.976/29.97 framerares. ???? Whats going on. Please help.

Gurgen

[mb1]

If you set "Multipass=3" you´re going to have 4 passes. First pass analyzes the video (.vaf) and then 3 passes to transcode the video.

CCE encodes NTSC sources always with 29.97 fps. Check the fps of your 2 files.

Don´t know Link2 and its abilities (maybe 3:2 pulldown integration ?).

[tyee]

@mb1
I've been doing the OPVBR lately like mikeathome. Do you know what CCE does here. I believe I can make a OPVBR without a .vaf file. The .vaf only is used on subsequent passes, which in this case there isn't any. So how is CCE making it VBR without using .vaf?

tyee

[gurgenpb]

mb1

I am not encoding an NTSC source. I am encoding a 23.976 fps film source. Also, this problem occurs when I use avisynth ONLY. With VFAPI avi file it works out great!!! (I later do pulldown in either case). I assign different file names for 1st and 2nd passes - that's how I know that the files are different in the avisynth's case (1st one of the correct size, 2nd one 80% of the desired size). When using VFAPI, both passes are of the same size. WEIRD!!!!

BTW: I thought that three passes means 1 pass for vaf file, and then 2 passes for the video file itself. Is this unnecessary? Are 2 passes enough for good quality video at 4-5 MPS bitrate?
What do u think is going on here!!!

[frank]

My experience is that 2 passes are sufficiently!
1. pass: analyzing with CBR.
If you don't change the bitrate in following passes then are all passes after the 2nd a great time wasting! Every following VBR pass calculates the average br in VAR and starts from this value. No visible improvements!

[gurgenpb]

thanks a lot
2 passes total!!!