Blender (non-proprietary Intermediate Codec) & PF24 pulldown

[avhell]

Hi all,

So, I have in the past used the mencoder settings from Eugenia's page to generate true 24P footage from my PF24 HV30 output.

However, those times did not include any A/V sync that I was concerned about. Now along a couple of years and I want to grab some more PF24 footage, but keep the sync, and I'm running into problems.

The mencoder settings have a really, really nasty a/v drift that shows up later in the output. (i'm using huffyuv as an intermediate codec, btw).

I was about to follow the suggestions to use neoscene to just apply the pulldown during capture, and tried the trial version out, but the problem is that to get the pulldown correct, you need to save to the Cineform HD codec (inside an .avi).

Which would be fine, but I am using Blender as my video editor... The builds I have seen do not have a way for me to read an proprietary codec like CFHD.

So - A huge cry out for help!

I've also tried Khaver's script, but I'm on Windows 7 64bit, and all of my output after following each of the instruction to the letter is a big fat video render of:


DirectShowSource: Could not open as video or audio.
... and more output saying the .m2t file "Cannot play back the file. The format is not supported."

I can only assume the problem is with ffdshow not decoding my file correctly, and feeding that bad output to avisynth (?).

So - please - it's 2011 for crying out loud! Is there a bulletproof way to convert this stinking mess of a PF24 file out to a true 24p file that is useful to me (in blender) w/o using a proprietary format?

In a nutshell -all of the free methods have produced either no useable output - or bad a/v sync issues.

The proprietary pay methods produce output that is locked into a format that I cannot use.

Any help?

TIA

[racer-x]

Assuming you have the latest FFdeshow installed:

You can use Virtualdub along with the buit-in IVTC plugin to remove Pulldown and export @ 23.976 fps to your available and compatible codecs or simply image sequence that I'm sure blender can handle.........

[avhell]

I have not been able to get ffdshow working correctly as far as I can tell.

What has worked for me is using avidemux, but I was wondering if I am setting the correct settings...

My filters in avidemux are:
1. Decomb Telecide (3:2 pulldown)
2. Decomb Decimate (Pulldown dupe removal)

The file seems to be ok (using ffvhuffyuv as intermediate) (even though they are chunked, and I can't find any way to make them a single large .avi file).

I haven't yet come across any nasty a/v sync issues as Eugenia reported on her page once upon a time...

[racer-x]

even though they are chunked, and I can't find any way to make them a single large .avi file

Are you using a FAT 32 formatted HDD and the avi is split into 4 GB chunks? You need to use NTFS if that's the case.

Your settings look right. If you are using an HDV camera, Virtualdub can import them directly if you rename the extension from .m2t to .ts instead. it's the same thing as both are Transport Streams. I forgot to mention that earlier.

[avhell]

racer-x: thank you for all the help, I truly appreciate it.

So it seems my problem is a non-issue at the moment (assuming my decomb telecide and decomb decimate are the correct options)?

On a side note, I know that the hdv standard has the image horizontally compressed to 1440, but i am wondering if just setting my output render (in blender for my case) to 1920 would be the correct way to handle editing this file? will the stretching be "correct" or is there a better method that should be employed?

[racer-x]

I believe blender 2.5 already has an HDV 1440x1080 preset. I don't see any harm in exporting at 1920x1080 as that will be the final viewed resolution anyway. If it were me I'd export at 1920x1080 when I'd remove pulldown and work with that resolution from then onward.